changes44.xml

<Chapter Label="HistoryOfGAP44"> 
<Heading>Overview of updates for &GAP; 4.4</Heading>

This chapter lists changes in the main system (excluding packages) 
that have been corrected or added in bugfixes and updates for 
&GAP; 4.4.

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 
<Section Label="fix442"> 
<Heading>&GAP; 4.4 Bugfix 2 (April 2004)</Heading>

Fixed bugs which could lead to crashes:
<List>              
<Item>
A crash when incorrect types of arguments are passed to <C>FileString</C>.
</Item>
</List>

Other fixed bugs:
<List>
<Item>A bug in <C>DerivedSubgroupTom</C> and <C>DerivedSubgroupsTom</C>.
</Item>
<Item>An error in the inversion of certain <C>ZmodnZObj</C> elements.
</Item>
<Item>A wrong display string of the numerator in rational 
functions returned by <C>MolienSeriesWithGivenDenominator</C> 
(in the case that the constant term of this numerator is zero).
</Item>
</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 

<Section Label="fix443">
<Heading>&GAP; 4.4 Bugfix 3 (May 2004)</Heading>

Fixed bugs which could produce wrong results:

<List>
<Item>Incorrect setting of system variables (e.g., home directory 
and command line options) after loading a workspace.
</Item>
<Item>Wrong handling of integer literals within functions or loops 
on 64-bit architectures (only integers in the range from 
<M>2^{28}</M> to <M>2^{60}</M>).
</Item>
</List>

Fixed bugs which could lead to crashes:

<List>
<Item>A problem in the installation of the multiplication routine 
for matrices that claimed to be applicable for more general list 
multiplications.
</Item>
<Item>A problem when computing weight distributions of codes with 
weights &gt; <M>2^{28}</M>.
</Item>
</List>

Other fixed bugs:

<List>
<Item>Problems with the online help with some manual sections.
</Item>
<Item>Problems of the online help on Windows systems.
</Item>
<Item>A problem in <C>GQuotients</C> when mapping from a finitely 
presented group which has a free direct factor.
</Item>
<Item>A bug in the function <C>DisplayRevision</C>.
</Item>
<Item>The trivial finitely presented group on no generators was not 
recognized as finitely presented.
</Item>
<Item>A problem with <C>Process</C>.
</Item>
<Item>A problem when intersecting subgroups of finitely presented 
groups that are represented in <Q>quotient representation</Q> with 
the quotient not apermutation group.
</Item>
<Item>A bug in the generic <C>Intersection2</C> method for vector 
spaces, in the case that both spaces are trivial.
</Item>
<Item>Enable ReeGroup(q) for <M>q = 3</M>.
</Item>
</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 

<Section Label="fix444">
<Heading>&GAP; 4.4 Bugfix 4 (December 2004)</Heading>

Fixed bugs which could produce wrong results:

<List>
<Item>
An error in the <C>Order</C> method for matrices over 
cyclotomic fields which caused this method to return <C>infinity</C> 
for matrices of finite order in certain cases.
</Item>
<Item>
Representations computed by <C>IrreducibleRepresentations</C>
in characteristic 0 erraneously claimed to be faithful. 
</Item>
<Item>A primitive representation of degree 574 for PSL(2,41) 
has been missing in the classification on which the &GAP; 
library was built. 
</Item>
<Item> 
A bug in <C>Append</C> for compressed vectors over GF(2): if the 
length of the result is 1 mod 32 (or 64) the last entry was 
forgotten to copy.
</Item>
<Item> 
A problem with the Ree group Ree(3) of size 1512 claiming 
to be simple.
</Item>
<Item>
An error in the membership test for groups 
GU(n,q) and SU(n,q) for non-prime <M>q</M>.
</Item>
<Item>
An error in the kernel code for ranges which caused e.g.
<C>-1 in [1..2]</C> to return <K>true</K>.
</Item>
<Item>
An error recording boolean lists in saved workspaces.
</Item>
<Item>
A problem in the selection function for primitive and
transitive groups if no degree is given.
</Item>
<Item>
<C>ReducedConfluentRewritingSystem</C> 
returning a cached result that might
not conform to the ordering specified.
</Item>
</List>

Other fixed bugs:

<List>
<Item> 
A problem with the function <C>SuggestUpdates</C> to check for 
the most recent version of packages available.
</Item>
<Item> 
A problem that caused <C>MatrixOfAction</C> to produce an error when
the algebra module was constructed as a direct sum.
</Item>
<Item> 
Problems with computing <M>n</M>-th power maps of character tables,
where n is negative and the table does not yet store its irreducible
characters.
</Item>
<Item>
Element conjugacy in large-base permutation groups sometimes was
unnecessarily inefficient.
</Item>
<Item>
A missing method for getting the letter representation of an associate
word in straight line program representation.
</Item>
<Item>
A problem with the construction of vector space bases where the given list 
of basis vectors is itself an object that was returned by <C>Basis</C>.
</Item>
<Item> 
A problem of <C>AbelianInvariantsMultiplier</C> insisting that a result of
<C>IsomorphismFpGroup</C> is known to be surjective.
</Item>
<Item> 
An error in the routine for <C>Resultant</C> 
if one of the polynomials has degree zero.
</Item>
</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 

<Section Label="fix445">
<Heading>&GAP; 4.4 Update 5 (May 2005)</Heading>

Fixed bugs which could produce wrong results:
<List>
<Item>
<Ref BookName="ref" Func="GroupWithGenerators"/> 
returned a meaningless group object instead of signaling an error
when it was called with an empty list of generators.
</Item>
<Item>
When computing preimages under an embedding into a direct product
of permutation groups, if the element was not in the image of the 
embedding then a permutation had been returned instead 
of <C>fail</C>.
</Item>
<Item>
Two problems with <Ref BookName="ref" Func="PowerMod"/> 
for polynomials. [Reported by Jack Schmidt]
</Item>
<Item>
Some methods for computing the sum of ideals returned the first 
summand instead of the sum. [Reported by Alexander Konovalov]
</Item>
<Item>
Wrong result in <Ref BookName="ref" Func="Intersection"/> for pc groups.
</Item>
<Item>
The function <Ref BookName="ref" Func="CompareVersionNumbers"/> 
erroneously ignored leading non-digit characters.
<P/>
A new feature in the corrected version is an optional third argument
<C>"equal"</C>, which causes the function to return
<C>true</C> only if the first two arguments describe
equal version numbers; documentation is available in the ext-manual.
This new feature is used in <Ref BookName="ref" Func="LoadPackage"/>,
now one can require a specific version of a package.
<P/>
The library code still contained parts of the handling of completion 
files for packages, which does not work and therefore had already 
been removed from the documentation. This code has now been removed.
<P/>
Now a new component <C>PreloadFile</C> is supported in
<Ref BookName="ref" Func="The PackageInfo.g File"/> files;
if it is bound then the file in question is read immediately before the
package or its documentation is loaded.
</Item>
<Item>
The result of <Ref BookName="ref" Func="String"/> for strings not 
in <Ref BookName="ref" Func="IsStringRep"/> that occur as list 
entries or record components was erroneously missing the double
quotes around the strings.
</Item>
<Item>
A bug which caused <Ref BookName="ref" Func="InducedPcgs"/> to 
return a pcgs which is not induced wrt. the parent pcgs of 
<C>pcgs</C>. This may cause unpredictable behaviour, e.&nbsp;g. 
when <C>SiftedPcElement</C> is used subsequently. 
[Reported by Alexander Konovalov]
</Item>
<Item>
Fixed a bug in <C>SmallGroupsInformation(512)</C>.
</Item>
<Item>
<Ref BookName="ref" Func="PowerModCoeffs"/> with exponent 1 for 
compressed vectors did not reduce (a copy of) the input vector 
before returning it. [Reported by Frank Lübeck]
</Item>
<Item>
Sorting a mutable non-plain list (e.g., a compressed matrix over 
fields of order &lt; 257) could potentially destroy that object. 
[Reported by Alexander Hulpke]
</Item>
<Item>
Under rare circumstances computing the closure of a permutation 
group by a normalizing element could produce a corrupted stabilizer 
chain. (The underlying algorithm uses random elements, probability 
of failure was below 1 percent). [Reported by Thomas Breuer]
</Item>
</List>

Fixed bugs which could lead to crashes:
<List>
<Item>
Some code and comments in the &GAP; kernel assumed that there is no 
garbage collection during the core printing function <C>Pr</C>, 
which is not correct. This could cause &GAP; in rare cases to 
crash during printing permutations, cyclotomics or strings with 
zero bytes. [Reported by Warwick Harvey]
</Item>
</List>

Other fixed bugs:
<List>

<Item>
A rare problem with the choice of prime in the Dixon-Schneider 
algorithm for computing the character table of a group. 
[Reported by Jack Schmidt]
</Item>

<Item>
<Ref BookName="ref" Func="DirectProduct"/> for trivial permutation 
groups returned a strange object.
</Item>

<Item>
A problem with <Ref BookName="ref" Func="PolynomialReduction"/> 
running into an infinite loop.
</Item>

<Item>
Adding linear mappings with different image domains 
was not possible. [Reported by Pasha Zusmanovich]
</Item>

<Item>
Multiplying group ring elements with rationals was not 
possible. [Reported by Laurent Bartholdi]
</Item>

<Item>
<Ref BookName="ref" Func="Random"/>
now works for finite fields of size
larger than <M>2^{28}</M>. [Reported by Jack Schmidt]
</Item>

<Item>
Univariate polynomial creators did modify the coefficient 
list passed. [Reported by Jürgen Müller]
</Item>

<Item>
Fixed <Ref BookName="ref" Func="IntHexString"/> to accept 
arguments not in <C>IsStringRep</C>; the argument is
now first converted if necessary. [Reported by Kenn Heinrich]
</Item>

<Item>
The library code for stabilizer chains contained quite some
explicit references to the identity <C>()</C>.
This is unfortunate if one works with permutation groups, the
elements of which are not plain permutations but objects which
carry additional information like a memory, how they were obtained
from the group generators. For such cases it is much cleaner to
use the <C>One(...)</C> operation instead of <C>()</C>, such that 
the library code can be used for a richer class of group objects. 
This fix contains only rather trivial changes <C>()</C> to 
<C>One(...)</C> which were carefully checked by me. The
tests for permutation groups all run without a problem. However, it is
relatively difficult to provide test code for this particular change,
since the "improvement" only shows up when one generates new group
objects. This is for example done in the package 
<Package>recog</Package> which is in preparation. 
[Reported by Akos Seress and Max Neunhöffer]
</Item>

<Item>
Using <C>{}</C> to select elements of a known inhomogenous dense list
produced a list that might falsely claim to be known inhomogenous,
which could lead to a segfault if the list typing code tried to mark
it homogenous, since the code intended to catch such errors also had 
a bug. [Reported by Steve Linton]
</Item>

<Item>
The record for the generic iterator construction of subspaces domains
of non-row spaces was not complete.
</Item>

<Item>
When a workspace has been created without packages(<C>-A</C> option) 
and is loaded into a &GAP; session without packages (same option)
then an error message is printed.
</Item>

<Item>
So far the functions <Ref BookName="ref" Func="IsPrimeInt"/> and 
<Ref BookName="ref" Func="IsProbablyPrimeInt"/> are essentially the
same except that <C>IsPrimeInt</C> issues an additional 
warning when (non-proven) probable primes are considered as primes.
<P/>
These warnings now print the probable primes in question as well;  
if a probable prime is used several times then the warning is also 
printed several times; there is no longer a warning for some known 
large primes; the warnings can be switched off. See 
<Ref BookName="ref" Func="IsPrimeInt"/> for more details.
<P/>
If we get a reasonable primality test in &GAP; we will change the 
definition of <C>IsPrimeInt</C> to do a proper test.
</Item>

<Item>
Corrected some names of primitive groups in degree 26. 
[Reported by Robert F. Bailey]
</Item>

</List>

New or improved functionality:
<List>

<Item>
Several changes for <Ref BookName="ref" Func="ConwayPolynomial"/>:
<List>
<Item> many new pre-computed polynomials </Item>
<Item> put data in several separate files (only read when needed) </Item>
<Item> added info on origins of pre-computed polynomials </Item>
<Item> improved performance of <Ref BookName="ref" Func="ConwayPolynomial"/>
and <Ref BookName="ref" Func="IsPrimitivePolynomial"/> for p &lt; 256 </Item>
<Item> improved documentation of <C>ConwayPolynomial</C> </Item>
<Item> added and documented new functions 
<Ref BookName="ref" Func="IsCheapConwayPolynomial"/> and
<Ref BookName="ref" Func="RandomPrimitivePolynomial"/>
</Item>
</List>
</Item>

<Item>
Added method for <Ref BookName="ref" Func="NormalBase"/>
for extensions of finite fields.
</Item>

<Item>
Added more help viewers for the HTML version of the documentation 
(firefox, mozilla, konqueror, w3m, safari).
</Item>

<Item>
New function <Ref BookName="ref" Func="ColorPrompt"/>. 
(Users of former versions of a <C>colorprompt.g</C> file: 
Now you just need a <C>ColorPrompt(true);</C> in your 
<C>.gaprc</C> file.)
</Item>

<Item>

Specialised kernel functions to support <Package>GUAVA</Package>
2.0. &GAP; will only load <Package>GUAVA</Package> 
in version at least 2.002 after this update.

</Item>

<Item>
Now there is a kernel function <C>CYC_LIST</C>
for converting a list of rationals into a cyclotomic,
without arithmetics overhead.
</Item>

<Item>
New functions 
<Ref BookName="ref" Func="ContinuedFractionExpansionOfRoot"/> and
<Ref BookName="ref" Func="ContinuedFractionApproximationOfRoot"/>
for computing continued fraction expansions and continued fraction 
approximations of real roots of polynomials with integer coefficients.
</Item>

<Item>
A method for computing structure descriptions for finite groups,
available via <Ref BookName="ref" Func="StructureDescription"/>.
</Item>

<Item>
This change contains the new, extended version of the 
<Package>SmallGroups</Package> package. For example, the groups of 
orders <M>p^4</M>, <M>p^5</M>, <M>p^6</M> for arbitrary primes 
<M>p</M>, the groups of square-free order and the groups of 
cube-free order at most 50000 are included now. For more detailed 
information see the announcement of the extended package. 
</Item>

<Item>
The function <C>ShowPackageVariables</C> gives
an overview of the global variables in a package.
It is thought as a utility for package authors and referees.
(It uses the new function <C>IsDocumentedVariable</C>.)
</Item>

<Item>
The mechanisms for testing &GAP; has been improved:
<List>
<Item> The information whether a test file belongs to the list in
  <C>tst/testall.g</C> is now stored in the test file
  itself.</Item>
<Item> Some targets for testing have been added to the
  <C>Makefile</C> in the &GAP; root directory,
  the output of the tests goes to the new directory
  <C>dev/log</C>.</Item>
<Item> Utility functions for testing are in the new file
  <C>tst/testutil.g</C>.
  Now the loops over (some or all) files <C>tst/*.tst</C>
  can be performed with a function call, and the file
  <C>tst/testall.g</C> can be created automatically;
  the file <C>tst/testfull.g</C> is now obsolete.
  The remormalization of the scaling factors can now be done using a &GAP;
  function, so the file <C>tst/renorm.g</C> is
  obsolete.</Item>
<Item> Now the functions <C>START_TEST</C>
  and <C>STOP_TEST</C> use components in 
  <C>GAPInfo</C> instead of own globals,
  and the random number generator is always reset in
  <C>START_TEST</C>.</Item>
<Item> <C>GAPInfo.SystemInformation</C> now takes two
  arguments, now one can use it easier in the tests.</Item>
</List>
</Item>

<Item>
<Ref BookName="ref" Func="MultiplicationTable"/> is now an attribute,
and the construction of a magma, monoid, etc. from multiplication 
tables has been unified.
</Item>
</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 

<Section Label="fix446">
<Heading>&GAP; 4.4 Update 6 (September 2005)</Heading>

Attribution  of bugfixes  and  improved functionalities  to those  who
reported or  provided these, respectively, is  still fairly incomplete
and inconsistent with this update. We apologise for this fact and will
discuss until the next update how to improve this feature.
<P/>

Fixed bugs which could produce wrong results:
<List>

<Item>
The perfect group library does not contain any information 
on the trivial group, so the trivial group must be handled 
specially. <Ref BookName="ref" Func="PerfectGroup"/> and
<C>NrPerfectLibraryGroups</C> were changed
to indicate that the trivial group is not part of the library.
</Item>

<Item>
The descriptions of <C>PerfectGroup(734832,3)</C>
and <C>PerfectGroup(864000,3)</C> were corrected
in the <Ref BookName="ref" Func="Finite Perfect Groups"/>
library of perfect groups.
</Item>

<Item>
The functions 
<Ref BookName="ref" Func="EpimorphismSchurCover"/> and
<Ref BookName="ref" Func="AbelianInvariantsMultiplier"/>
may have produced wrong results without warning
[Reported by Colin Ingalls]. These problems are fixed.
However, the methods currently used can be expected to 
be slower than the ones used before; we hope to fix this 
in the next version of &GAP;.
</Item>

<Item>
<Ref BookName="ref" Func="DerivedSubgroup"/> and 
<Ref BookName="ref" Func="CommutatorSubgroup"/> for permutation 
groups sometimes returned groups with an incorrect stabilizer 
chain due to a missing verification step after a random Schreier Sims.
</Item>

<Item>
<Ref BookName="ref" Func="NaturalHomomorphismByNormalSubgroup"/>
for FpGroups did unnecessary rewrites.
</Item>

<Item>
The alternating group <M>A_3</M> incorrectly claimed to be not simple.
</Item>

<Item>
<Ref BookName="ref" Func="ExponentSyllable"/> for straight line 
program elements gave a wrong result.

</Item>

<Item>
<Ref BookName="ref" Func="PrimePGroup"/> is defined to return 
<C>fail</C> for trivial groups, but if the group was constructed 
as a factor or subgroup of a known <M>p</M>-group, the value of 
<M>p</M> was retained.
</Item>

<Item>
The functions <Ref BookName="ref" Func="TestPackageAvailability"/>
and <Ref BookName="ref" Func="LoadPackage"/> did not work correctly 
when one asked for a 
particular version of the package, via a version number starting with
the character <C>=</C>, in the sense that a version
with a larger version number was loaded if it was available. 
[Reported by Burkhard Höfling]
</Item>

<Item>
The generator names constructed by
<Ref BookName="ref" Func="AlgebraByStructureConstants"/> 
were nonsense.
</Item>

<Item>
The undocumented function (but recently advertised on gap-dev) 
<C>COPY_LIST_ENTRIES</C> did not handle overlapping source and
destination areas correctly in some cases.
</Item>

<Item>
The elements in a free magma ring have the filter
<Ref BookName="ref" Func="IsAssociativeElement"/> set
whenever the elements in the underlying magma and in 
the coefficients ring have this filter set. 
[Reported by Randy Cone]
</Item>

<Item>
The function <Ref BookName="ref" Func="InstallValue"/> must not 
be used for objects in the filter <C>IsFamily</C> because these
objects are compared via <Ref BookName="ref" Func="IsIdenticalObj"/>. 
[Reported by Max Neunhöffer]
</Item>

</List>

Fixed bugs which could lead to crashes:
<List>

<Item>
Problem in composition series for permutation groups for 
non-Frobenius groups with regular point stabilizer.
</Item>

<Item>
After lots of computations with compressed GF(2) vectors 
&GAP; occasionally crashed. The reason were three missing 
<C>CHANGED_BAG</C>s in <C>SemiEchelonPListGF2Vecs</C>.
They were missing, because a garbage collection could be 
triggered during the computation such that newly created bags 
could become <Q>old</Q>. It is not possible to provide test 
code because the error condition cannot easily be reproduced. 
[Reported by Klaus Lux]
</Item>

<Item>
Minor bug that crashed &GAP;:
The type of <C>IMPLICATIONS</C>
could not be determined in a fresh session. [Reported by Marco Costantini]
</Item>

<Item>
<Ref BookName="ref" Func="Assert"/>
caused an infinite loop if called
as the first line of a function called from another function. 
</Item>
</List>

Other fixed bugs:
<List>

<Item>
Wrong choice of prime in Dixon-Schneider if prime is bigger than 
group order (if group has large exponent).
</Item>

<Item>
Groebner basis code ran into problems when comparing monomial 
orderings.
</Item>

<Item>
When testing for conjugacy of a primitive group to an imprimitive 
group,&GAP; runs into an error in EARNS calculation. 
[Reported by John Jones]
</Item>

<Item>
The centre of a magma is commonly defined to be the set of elements 
that commute and associate with all elements. The previous definition
left out <Q>associate</Q> and caused problems with extending the 
functionality to nonassociative loops. [Reported by Petr Vojtechovsky]
</Item>

<Item>
New kernel methods for taking the intersection and difference between
sets of substantially different sizes give a big performance increase.
</Item>

<Item>
The commands
<Ref BookName="ref" Func="IsNaturalSymmetricGroup"/> and
<Ref BookName="ref" Func="IsNaturalAlternatingGroup"/> are faster and 
should run much less
often into inefficient tests.
</Item>

<Item>
The perfect group library, see 
<Ref BookName="ref" Func="Finite Perfect Groups"/>, is split into 
several files which are loaded and unloaded to keep memory usage 
down. The global variable <C>PERFSELECT</C> is a blist which
indicates which orders are currently loaded. An off-by-one error 
wrongly added the last order of the previous file into the list of 
valid orders when a new file was loaded. A subsequent access to 
this order raises an error.
</Item>

<Item>
Up to now, the method installed for testing the membership of rationals
in the field of rationals via <Ref BookName="ref" Func="IsRat"/>
was not called; instead a more general method was used that called
<Ref BookName="ref" Func="Conductor"/> and thus was much slower.
Now the special method has been ranked up by changing the requirements
in the method installation.
</Item>

<Item>
Fixed a bug in <C>APPEND_VEC8BIT</C>, which was triggered in the 
following situation: Let <C>e</C> be the number of field elements 
stored in one byte. If a compressed 8bit-vector <C>v</C> had length 
not divisible by <C>e</C> and another compressed 8-bit vector 
<C>w</C> was appended, such that the sum of the lengths became 
divisible by <C>e</C>, then one 0 byte too much was written, which 
destroyed the <C>TNUM</C> of the next &GAP; object in memory. 
[Reported by Klaus Lux]
</Item>

<Item>
<Ref BookName="ref" Func="PermutationCycle"/> returned <C>fail</C> 
if the cycle was not a contiguous subset of the specified domain. 
[Reported by Luc Teirlinck]
</Item>

<Item>
Now <Ref BookName="ref" Func="Inverse"/> correctly returns 
<C>fail</C> for zeros in finite fields (and does no longer 
enter a break loop).
</Item>

<Item>
Up to now, <Ref BookName="ref" Func="CharacterDegrees"/> ignored the attribute
<Ref BookName="ref" Func="Irr"/>
if the argument was a group that knew that it was solvable.
</Item>

<Item>
The function <C>Debug</C> now prints a meaningful message
if the user tries to debug an operation.
Also, the help file for <C>vi</C> is now available in
the case of several &GAP; root directories.
</Item>

<Item>
It is no longer possible to create corrupt objects via ranges of 
length &gt;<M>2^{28}</M>, resp. &gt;<M>2^{60}</M> (depending on 
the architecture). The limitation concerning the arguments of 
ranges is documented. [Reported by Stefan Kohl]
</Item>

<Item>
Now <Ref BookName="ref" Func="IsElementaryAbelian"/> and
<Ref BookName="ref" Func="ClassPositionsOfMinimalNormalSubgroups"/>
are available for ordinary character tables. Now the operation
<Ref BookName="ref" Func="CharacterTableIsoclinic"/> is 
an attribute, and there is another new attribute
<Ref BookName="ref" Func="SourceOfIsoclinicTable"/> that points back to the
original table; this is used for computing the Brauer tables of those 
tables in the character table library that are computed using
<C>CharacterTableIsoclinic</C>.
Now <Ref BookName="ref" Func="ClassPositionsOfDerivedSubgroup"/> 
avoids calling <Ref BookName="ref" Func="Irr"/>,
since <Ref BookName="ref" Func="LinearCharacters"/> is sufficient.
Now <Ref BookName="ref" Func="ClassPositionsOfElementaryAbelianSeries"/>
works also for the table of the trivial group.
Restrictions of character objects know that they are characters.
<P/>
A few formulations in the documentation concerning character tables have
been improved slightly.
</Item>

<Item>
Up to now, <Ref BookName="ref" Func="IsPGroup"/> has rarely been set. 
Now many basic operations such as <Ref BookName="ref" Func="SylowSubgroup"/> 
set this attribute on the returned result.
</Item>

<Item>
Computing an enumerator for a semigroup required too much time
because it used all elements instead of the given generators. 
[Reported by Manuel Delgado]
</Item>

<Item>
Avoid potential error message when working with automorphism groups.
</Item>

<Item>
Fixed wrong page references in manual indices.
</Item>

<Item>
Make <C>MutableCopyMat</C> an operation and install 
the former function which does call <Ref BookName="ref" Func="List"/> with
<Ref BookName="ref" Func="ShallowCopy"/> the default method for lists. Also
use this in a few appropriate places.
</Item>

<Item>
An old DEC compiler doesn't like C preprocessor directives that are 
preceded by whitespace. Removed such whitespace. [Reported by Chris Wensley]
</Item>

</List>


New or improved functionality:

<List>

<Item>
The primitive groups library has been extended to degree 2499.
</Item>

<Item>
New operation 
<Ref BookName="ref" Func="Remove"/>
and extended functionality of
<Ref BookName="ref" Func="Add"/> with
an optional argument giving the position of the insertion. They are based
on an efficient kernel function <C>COPY_LIST_ENTRIES</C>.
</Item>

<Item>
Added fast kernel implementation of Tarjan's 
algorithm for strongly connected components of a 
directed graph. 
</Item>

<Item>
Now <Ref BookName="ref" Func="IsProbablyPrimeInt"/> can be used with larger numbers. 
(Made internal function <C>TraceModQF</C> non-recursive.)
</Item>

<Item>
A new operation <Ref BookName="ref" Func="PadicValuation"/> 
and a corresponding method for rationals.
</Item>

<Item>
A new operation <Ref BookName="ref" Func="PartialFactorization"/> has been
added, and a corresponding method for integers has been installed.
This method allows one to specify the amount of work to be spent on looking
for factors.
</Item>

<Item>
The generators of full s. c. algebras can now be accessed 
with the dot operator. [Reported by Marcus Bishop]
</Item>

<Item>
New Conway polynomials computed by Kate Minola, John Bray, Richard Parker.
</Item>

<Item>
A new attribute <Ref BookName="ref" Func="EpimorphismFromFreeGroup"/>.
The code has been written by Alexander Hulpke.
</Item>

<Item>
The functions
<Ref BookName="ref" Func="Lambda"/>,
<Ref BookName="ref" Func="Phi"/>,
<Ref BookName="ref" Func="Sigma"/>, and
<Ref BookName="ref" Func="Tau"/>
have been turned into operations, to admit the installation of methods
for arguments other than integers.
</Item>

<Item>
Up to now, one could assign only lists with
<Ref BookName="ref" Func="InstallFlushableValue"/>.
Now also records are admitted.
</Item>

<Item>
<Ref BookName="ref" Func="InstallMethod"/> now admits entering a list of
strings instead of a list of required filters.
Each such string must evaluate to a filter when used as the argument of
<Ref BookName="ref" Func="EvalString"/>.
The advantage of this variant is that these strings are used to compose
an info string (which is shown by <C>ApplicableMethod</C>)
that reflects exactly the required filters.
</Item>

<Item>
In test files that are read with <C>ReadTest</C>,
the assertion level is set to 2 between <C>START_TEST</C>
and <C>STOP_TEST</C>.
This may result in runtimes for the tests that are substantially longer
than the usual runtimes with default assertion level 0.
In particular this is the reason why some of the standard test files require
more time in &GAP; 4.4.6 than in
&GAP; 4.4.5.
</Item>

<Item>
Some very basic functionality for floats.
</Item>

</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 

<Section Label="fix447">
<Heading>&GAP; 4.4 Update 7 (March 2006)</Heading>

New or improved functionality:

<List>

<Item>
The <Ref BookName="ref" Func="Display"/>
 functionality for character tables has been extended by addition 
of an option to show power maps and centralizer orders in a format
similar to that used in the ATLAS.
Furthermore the options handling is now hierarchical,
in order to admit more flexible overloading.
</Item>

<Item>
For the function <Ref BookName="ref" Func="LowIndexSubgroupsFpGroup"/>,
there is now an iterator variant
<C>LowIndexSubgroupsFpGroupIterator</C>.
[Suggested (and based on code contributed) by Michael Hartley]
</Item>

<Item>
Semigroup functionality in &GAP; has been improved 
and extended. Green's relations are now stored differently, making the system 
more amenable to new methods for computing these relations in special cases. 
It is now possible to calculate 
Green's classes etc. without computing the entire semigroup or necessarily 
loading the package <Package>MONOID</Package>. 
Furthermore, the Froidure-Pin algorithm has now been implemented in 
&GAP;. 
</Item>

<Item>
Functionality for creating free products of any list of groups for which 
a finite presentation can be determined had been added. This
function returns a finitely presented group. This functionality includes
the <C>Embedding</C> operation.

As an application of this new code a specialized direct product 
operation has been added for finitely presented groups which 
returns a finitely presented group. This application includes 
<C>Embedding</C> and <C>Projection</C> functionality.
</Item>

<Item>
Some new Straight Line Program (SLP) functionality has been added. 
The new functions take given SLPs and create new ones by restricting
to a subset of the results, or to an intermediate result or by calculating
the product of the results of two SLPs.
</Item>

<Item>
New code has been added to allow group elements with memory;
that is, they store
automatically how they were derived from some given set of generators.
Note that there is not yet documentation for this functionality,
but some packages already use it.
</Item>

<Item>
New code has been added to handle matrices and vectors in such a way that 
they do not change their representation in a generic manner.
</Item>

<Item>
The <Ref BookName="ref" Func="Irr"/>
method for <M>p</M>-solvable <M>p</M>-modular Brauer tables now keeps 
the order of the irreducibles in the ordinary table.
</Item>

<Item>
&GAP; can now handle any finite field for which the 
Conway polynomial is known or can be computed. 
</Item>

<Item>
New Conway polynomials provided by John Bray and Kate Minola have been added.
</Item>

<Item>
The <C>ReadTest</C>
 methods for strings (filenames) and streams now automatically set
the screen width (see <Ref BookName="ref" Func="SizeScreen"/>) to 80 before the tests,
and reset it afterwards.
</Item>

<Item>
Now a few more checks are done during the <C>configure</C>
phase of compiling for future use of some I/O functions of the C-library
in a package. Also the path to the &GAP; binaries for the &GAP; compiler is 
now handled via autoconf. Finally, now <C>autoconf</C>
version 2.59 is used.
</Item>

</List>

Fixed bugs which could produce wrong results:

<List>

<Item>
Some technical errors in the functions for compressed vectors and matrices
which could lead to corruption of internal data structures and so to crashes
or conceivably to wrong results. [Reported by Roman Schmied]
</Item>

<Item>
A potential problem in the generic method for the undocumented operation
<C>DirectFactorsOfGroup</C>:
It was silently assumed that
<Ref BookName="ref" Func="NormalSubgroups"/>
delivers the trivial subgroup as first and
the whole group as last entry of the resulting list.
</Item>

<Item>
The code for sublists of compressed vectors created by <C>vec{range}</C> may write 
one byte beyond the space allocated for the new vector, overwriting part of the next
object in the workspace. Thanks to Jack Schmidt for narrowing down the problem.
</Item>

<Item>
Given a class function object of value zero, an 
<Ref BookName="ref" Func="Arithmetic Operations for Class Functions"/> 
method for a class function
erroneously did not return <C>fail</C>. 
[Reported by Jack Schmidt]
</Item>

<Item>
The <Ref BookName="ref" Func="Arithmetic Operations for Class Functions"/> 
method for a class function
erroneously returned a finite number if one of the values
was nonreal, not a cyclotomic integer, and had norm 1.
</Item>

<Item>
Two missing perfect groups were added, and the permutation degree lowered on
the perfect groups with the largest degrees. [Reported by Jack Schmidt]
</Item>

<Item>
When a character table was displayed with
<Ref BookName="ref" Func="Printing Character Tables"/>,
the centralizer order displayed for the first class shown
was not correct if it did not involve
all prime divisors of the group. [Reported by Jack Schmidt]
</Item>

<Item>
The first argument of the function <Ref BookName="ref" Func="VectorSpace"/> must be a field.
This is checked from now on. [Reported by Laurent Bartholdi]
</Item>

<Item>
Up to now, it was possible to create a group object from a semigroup of
cyclotomics using <Ref BookName="ref" Func="AsGroup"/>,
although groups of cyclotomics are not admissible. [Reported by Alexander Konovalov]
</Item>

<Item>
The documentation of 
<C>CharacteristicPolynomial(F,mat)</C> 
was ambiguous if
<C>FieldOfMatrix(mat) &lt;= F &lt; 
DefaultFieldOfMatrix(mat)</C>.
In particular, the
result was representation dependent. This was fixed by introducing a second
field which specifies the vector space which mat acts upon.  [Reported by Jack Schmidt]
</Item>

<Item>
<C>AssociatedReesMatrixSemigroupOfDClass</C> produced an incorrect 
sandwich matrix for the semigroup created. This matrix is an
attribute set when creating the Rees matrix semigroup but is 
not used for creating the semigroup. The incorrect result was returned
when <C>SandwichMatrix</C> was called. [Reported by Nelson Silva and Joao Araujo]
</Item>

<Item>
The literal <C>"compiled"</C> was given an incorrect length. The 
kernel was then unable to find compiled library code as the search
path was incorrect. Also the documentation example had an error in 
the path used to invoke the <F>gac</F> compiler.
</Item>

<Item>
The twisting group in a generic wreath product might
have had intransitive action. [Reported by Laurent Bartholdi]
</Item>

<Item>
There was an arithmetic bug in the polynomial reduction code.
</Item>
</List>


Fixed bugs which could lead to crashes:

<List>

<Item>
Bug 1 in the list of fixed bugs which could lead to wrong results
could also potentially lead to crashes.
</Item>

</List>


Other fixed bugs:

<List>

<Item>
The matrices of invariant forms stored as values of the attributes
<Ref BookName="ref" Func="InvariantBilinearForm"/>,
<Ref BookName="ref" Func="InvariantQuadraticForm"/>, and
<Ref BookName="ref" Func="InvariantSesquilinearForm"/>,
for matrix groups over finite fields,
are now in the (compressed) format returned by
<Ref BookName="ref" Func="ImmutableMatrix"/>.
</Item>

<Item>
<C>String</C> now returns an immutable string, 
by making a copy before changing the argument. 
</Item>

<Item>
<C>permutation^0</C> and <C>permutation^1</C> were not handled with 
special code in the kernel, hence were very slow for big permutations.
[Reported by Max Neunhöffer]
</Item>

<Item>
Added code to cache the induced pcgs for an arbitrary parent pcgs. (This code
was formerly part of the <Package>CRISP</Package> package.)
</Item>

<Item>
This fix consists of numerous changes to improve support for 
direct products, including:
- new methods for <C>PcgsElementaryAbelianSeries</C>,
<C>PcgsChiefSeries</C>,
<C>ExponentsOfPcElement</C>, 
<C>DepthOfPcElement</C> for direct products
- fixed <C>EnumeratorOfPcgs</C> to test for membership first
- new methods for membership test in groups which have an induced pcgs
- added <C>GroupOfPcgs</C> attribute to pcgs in various methods
- fixed declarations of <C>PcgsElementaryAbelianSeries</C>, 
<C>PcgsChiefSeries</C>
(the declared argument was a pcgs, not a group) [Reported by Roman Schmied]
</Item>

<Item>
Corrected a term ordering problem encountered by the basis construction
code for finite dimensional vector spaces of multivariate rational 
functions. [Reported by Jan Draisma]
</Item>

<Item>
When the factor of a finite dimensional group ring by an ideal was formed,
a method intended for free algebras modulo relations was used,
and the returned factor algebra could be used for (almost) nothing. 
[Reported by Heiko Dietrich]
</Item>

<Item>
Up to now, <Ref BookName="ref" Func="PowerMap"/> ran into an error
when one asked for the n-th power map where n was not a small integer.
This happened in some &GAP; library functions
if the exponent of the character table in question was not a small integer.
</Item>

<Item>
Up to now, the test whether a finite field element was contained in a group
of finite field elements ran into an error if the element was not in the
field generated by the group elements. [Reported by Heiko Dietrich]
</Item>

<Item>
Conjugacy classes of natural (special) linear groups are now always returned
with trivial class first.
</Item>

<Item>
Up to now, it could happen that <Ref BookName="ref" Func="CheckFixedPoints"/> 
reduced an entry in its
second argument to a list containing only one integer
but did not replace the list by that integer;
according to the conventions, this replacement should be done.
</Item>

<Item> 
The functions <C>PrintTo</C> and 
<C>AppendTo</C> did not work correctly for streams.
[Reported by Marco Costantini]
</Item>

<Item>
The function <C>Basis</C> did not return a value
when it was called with the argument <C>Rationals</C>.
[Reported by Klaus Lux]
</Item>

<Item>
For certain matrix groups, the function
<C>StructureDescription</C> raised an error message.
The reason for this was that a trivial method for
<C>IsGeneralLinearGroup</C> for matrix groups in
<C>lib/grpmat.gi</C> which is ranked higher than
the nontrivial method for generic groups in
<C>lib/grpnames.gi</C> called the operation
<C>IsNaturalGL</C>, for which there was no nontrivial
method available. [Reported by Nilo de Roock]
</Item>

<Item> 
Action on sets of length 1 was not correctly handled. 
[Reported by Mathieu Dutour]
</Item>

<Item>
Now <C>WriteByte</C> admits writing zero characters to
all streams. [Reported by Marco Costantini]
</Item>

<Item>
The conjugacy test for subgroups tests for elementary abelian regular 
normal subgroup (EARNS) conjugacy. The fix will 
catch this in the case that the second group has no EARNS. 
[Reported by Andrew Johnson]
</Item>

<Item>
So far, the UNIX installation didn't result in a correct gap.sh if the 
installation path contained space characters. Now it should handle this 
case correctly, as well as other unusual characters in path names 
(except for double quotes).
</Item>

</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 

<Section Label="fix448">
<Heading>&GAP; 4.4 Update 8 (September 2006)</Heading>

New or improved functionality:

<List>

<Item>
A function <Ref BookName="ref" Func="Positions"/> with underlying operation
<C>PositionsOp</C>, which returns the list
of all positions at which a given object appears in a given list.
</Item>

<Item>
<Ref BookName="ref" Func="LogFFE"/> 
now returns  <C>fail</C> 
when the element is not a power of the base.
</Item>

<Item>
It is now allowed to continue long integers, strings or identifiers by
ending a line with a backslash or with a backslash and carriage return
character. So, files with &GAP; code and 
DOS/Windows-style  line breaks  are now valid input on all architectures.
</Item>

<Item>
The command line for starting the session and the system environment are now
available in <C>GAPInfo.SystemCommandLine</C> and
<C>GAPInfo.SystemEnvironment</C>.
</Item>

<Item>
Names of all bound global variables and all component names are available 
on &GAP; level.
</Item>

<Item>
Added a few new Conway polynomials computed by Kate Minola and John Bray.
</Item>

<Item>
There is a new concept of <E>random sources</E>, see 
<Ref BookName="ref" Func="IsRandomSource"/>, which provides random number 
generators which are independent of each other.
There is kernel code for the Mersenne twister random number generator
(based on the code by Makoto Matsumoto distributed at
<URL>http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/emt.html</URL>).
It provides fast 32-bit pseudorandom integers with a period of length 
<M>2^{19937}-1</M>
and a 623-dimensional equidistribution. The library methods for random
elements of lists and for random (long) integers are using the Mersenne 
twister now.<P/>
</Item>

<Item>
In line editing mode (usual input mode without -n option) in lines starting
with <C>gap&gt; </C>, <C>&gt; </C> or
<C>brk&gt; </C> this beginning part is immediately
removed. This is a convenient feature that allows one to cut and paste
input lines from other sessions or from manual examples into the current 
session.
</Item>
</List>

Fixed bugs which could produce wrong results:

<List>

<Item>
The function <Ref BookName="ref" Func="Decomposition"/> 
returned coefficient vectors also in certain
situations where in fact no decomposition exists.
This happened only if the matrix entered as the first argument
contained irrational values
and a row in the matrix entered as the second argument did not respect the
algebraic conjugacy relations between the columns of the first argument.
So there was no problem for the usual cases that the two matrices are
integral or that they are lists of Brauer characters. [Reported by Jürgen Müller]
</Item>

<Item>
PC group homomorphisms can claim a wrong kernel after 
composition. [Reported by Serge Bouc]
</Item>

<Item>
The return value of <Ref BookName="ref" Func="OctaveAlgebra"/> had an inconsistent
defining structure constants table for the case of coefficients fields
not containing the integer zero. [Reported by Gábor Nagy]
</Item>

<Item>
The manual guarantees that a conjugator automorphism has a conjugating
element in the group if possible. This was not guaranteed.
</Item>

<Item>
<Ref BookName="ref" Func="StabChain"/>
for symmetric groups gave a wrong result if fixed points were
prescribed for base.
</Item>

<Item>
Contrary to what is documented the function 
<C>POW_OBJ_INT</C> returned an immutable result for
<C>POW_OBJ_INT(m,1)</C> for a mutable object
<C>m</C>. This is triggered by the code
<C>m^1</C>.
</Item>

<Item>
<Ref BookName="ref" Func="PseudoRandom"/> for a group 
had a problem if the group had lots of equal
generators. The produced elements were extremely poorly distributed in 
that case. This is now fixed for the case that elements of the group
can easily be sorted.
</Item>

<Item>
Fixed the bug that the type of a boolean list (see 
<Ref BookName="ref" Func="More about Boolean Lists"/>)
was computed wrongly: The type previously had <C>IS_PLIST_REP</C> 
instead of <C>IS_BLIST_REP</C> in its filter list.
</Item>

<Item>
<Ref BookName="ref" Func="Orbits"/> 
did not respect a special <Ref BookName="ref" Func="PositionCanonical"/> method for right
transversals. [Reported by Steve Costenoble]
</Item>

<Item>
Wrong results for <Ref BookName="ref" Func="GcdInt"/> 
for some arguments on 64 bit systems only.
 [Reported by Robert Morse]
</Item>

<Item>
When prescribing a subgroup to be included, the low index algorithm for fp
groups sometimes returned subgroups which are in fact conjugate. (No
subgroups are missing.) [Reported by Ignaz Soroko]
</Item>
</List>


Fixed bugs which could lead to crashes:

<List>

<Item>
The command line option <C>-x</C> allowed 
arguments > 256 which
can then result in internal buffers overflowing. Now bigger numbers
in the argument are equivalent to <C>-x 256</C>.
<!-- 
Can't really test it except by starting gap -x 300 and checking SizeScreen();
With the fix it is [256,24], without it was [300,24]. -->
[Reported by Michael Hartley]
</Item>
</List>


Other fixed bugs:

<List>

<Item>
Two special methods for the operation
<Ref BookName="ref" Func="CompositionMapping2"/> were not correct,
such that composing (and multiplying) certain group homomorphisms
did not work. [Reported by Peter Mayr]
</Item>

<Item>
In the definition of
<Ref BookName="ref" Func="FrobeniusCharacterValue"/>, it had been stated erroneously
that the value must lie in the field of <M>p^n</M>-th roots of unity;
the correct condition is that the value must lie in the field of
<M>(p^n-1)</M>-th roots of unity. [Reported by Jack Schmidt]
</Item>

<Item>
The function <Ref BookName="ref" Func="DirectProduct"/> failed when one of the factors was known to be
infinite.
</Item>

<Item>
For a linear action homomorphism <C>PreImageElm</C>
was very slow because there was no good method to check for injectivity,
which is needed for nearly all good methods for 
<C>PreImageElm</C>. This change adds such a new
method for <C>IsInjective</C>. [Reported by Akos Seress]
</Item>

<Item>
Rare errors in the complement routine for permutation groups.
</Item>

<Item>
Blocks code now uses jellyfish-style random elements to avoid bad Schreier
trees.
</Item>

<Item>
A method for <Ref BookName="ref" Func="IsPolycyclicGroup"/> has been added.
Such a method was missing so far.
</Item>

<Item>
Corrected <Ref BookName="ref" Func="EpimorphismSchurCover"/> to handle
the trivial group correctly.
Added new methods that follow immediately from computing 
the Schur Cover of a group. The attribute 
<Ref BookName="ref" Func="Epicentre"/>, the operations
<Ref BookName="ref" Func="NonabelianExteriorSquare"/> and
<Ref BookName="ref" Func="EpimorphismNonabelianExteriorSquare"/>, and
the property <Ref BookName="ref" Func="IsCentralFactor"/> are added to the 
library with documentation and references.
</Item>

<Item>
Display the correct expression in a call stack trace
if an operation was called somewhere up due to the evaluation
of a unary or binary operation.
</Item>

<Item>
Made <C>StripMemory</C> an operation rather than a 
global function. Added <C>ForgetMemory</C> operation.
</Item>

<Item>
Adjust things slightly to make later conversion to new vectors/matrices
easier. Nothing of this should be visible.
</Item>

<Item>
Corrected some details in the documentation of the &GAP; 
language. [Reported by Alexander Konovalov]
</Item>

<Item>
Now <Ref BookName="ref" Func="PositionSorted"/> is much faster on long mutable 
plain lists. (The former operation is substituted by a function and a new
operation <C>PositionSortedOp</C>.) [Reported by Silviu Radu]
</Item>

<Item>
Now it is possible to switch repeated warnings off when working with
iterative polynomial rings.
</Item>
</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 

<Section Label="fix449">
<Heading>&GAP; 4.4 Update 9 (November 2006)</Heading>

Fixed bugs which could produce wrong results:

<List>

<Item>
The methods of
<Ref BookName="ref" Func="ReadByte"/>
for reading from files or  terminals returned wrong
results for characters in the range <C>[128..255]</C>. 
[Reported by Yevgen Muntyan]
</Item>
</List>


Other fixed bugs:

<List>

<Item>
A method for the operation
<Ref BookName="ref" Func="PseudoRandom"/> did not succeed.
</Item>

<Item>
A fix for <C>Orbits</C> with a set of points as a seed.
</Item>

<Item>
Added a generic method such that <Ref BookName="ref" Func="Positions"/> 
works with all types of lists.
</Item>

<Item>
Fixed a problem in choosing the prime in the Dixon-Schneider algorithm. 
[Reported by Toshio Sumi]
</Item>
</List>

New or improved functionality:

<List>

<Item>
<C>ReducedOrdinary</C> was used in the manual, but was not documented, 
being a synonym for the documented <C>ReducedCharacters</C>. Changed 
manual examples to use the latter form. [Reported by Vahid Dabbaghian]
</Item>
</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 

<Section Label="fix4410">
<Heading>&GAP; 4.4 Update 10 (October 2007)</Heading>


New or improved functionality:

<List>

<Item>
Files in the <F>cnf</F> directory of the &GAP;
distribution are now archived as binary files. Now &GAP;
can be installed with UNIX or with WINDOWS style
line breaks on any system and should work without problems.
</Item>

<Item>
Since large finite fields are available,
some restrictions in the code for computing irreducible modules
over finite fields are no longer necessary.
(They had been introduced in order to give better error messages.)
</Item>

<Item>
Made PositionSublist faster in case the search string does not contain
repetitive patterns.
</Item>

<Item>
The function <C>MakeImmutable</C> now returns
its argument.
</Item>

<Item>
Dynamically loaded modules now work on Mac OS X. As a consequence,
this allows to work with the Browse, EDIM and IO packages on Mac OS X.
</Item>

<Item>
Introduced <C>ViewObj</C> and
<C>PrintObj</C> methods for algebraic number fields. Made
them applicable to <C>AlgebraicExtension</C> by adding the
property <C>IsNumberField</C> in the infinite field case.
</Item>

<Item>
The function
<Ref BookName="ref" Func="CharacterTableRegular"/>
is documented now.
</Item>

<Item>
The function
<Ref BookName="ref" Func="ScalarProduct"/>
now accepts also Brauer characters as arguments.
</Item>

<Item>
The function
<Ref BookName="ref" Func="QuaternionAlgebra"/>
now accepts also a list of field elements instead of a field.
Also, now the comparison of return values (w.r.t. equality, containment)
yields <C>true</C> if the parameters coincide and the
ground fields fit.
</Item>

<Item>
The function
<Ref BookName="ref" Func="RemoveCharacters"/> is now documented.
</Item>

<Item>
Lists in &GAP; sometimes occupy memory for
possible additional entries. Now plain lists and strings read by &GAP; 
and the lists returned by <Ref BookName="ref" Func="List"/>
only occupy the memory they really need. For more details see the
documentation of the new function 
<Ref BookName="ref" Func="EmptyPlist"/>.
</Item>

<Item>
There are some new Conway polynomials in characteristic 2 and 3 provided by 
Kate Minola.
</Item>

<Item>
A new operation <C>MemoryUsage</C> determines the
memory usage in bytes of an object and all its subobjects. It does
not consider families and types but handles arbitrary self-referential
structures of objects.
</Item>
</List>


Fixed bugs which could produce wrong results:

<List>

<Item>
When forming the semidirect product of a matrix group with a vector space
over a non-prime field
the embedding of the vector space gave a wrong result. [Reported by anvita21]
</Item>

<Item>
DefaultRing failed for constant polynomials over nonprime fields. 
[Reported by Stefan Kohl]
</Item>

<Item>
The method in ffeconway.gi that gets coefficients WRT to the 
canonical basis of the field from the representation is only correct
if the basis is over the prime field. Added a TryNextMethod if this 
is not the case. [Reported by Alla Detinko]
</Item>

<Item>

Creating a large (><M>2^{16}</M>) field over a non-prime subfield went 
completely wrong. [Reported by Jack Schmidt, from Alla Detinko]
</Item>

<Item>
A method for Coefficients for Conway polynomial FFEs 
didn't check that the basis provided was the canonical basis of
the RIGHT field. 
 [Reported by Bettina Eick]
</Item>

<Item>
An elementary abelian series was calculated wrongly. [Reported by N. Sieben]
</Item>

<Item>
Orbits on sets of transformations failed.
</Item>

<Item>
Wrong methods for <Ref BookName="ref" Func="GeneratorsOfRing"/>
and <Ref BookName="ref" Func="GeneratorsOfRingWithOne"/>
have been removed. These methods were based on the assumption
that one can obtain a set of ring generators by taking the
union of a known set of field generators, the set of the
inverses of these field generators and&nbsp;{1}.
</Item>

<Item>
The name of a group of order 117600 and degree 50 was
incorrect in the <Ref BookName="ref" Func="Primitive Permutation Groups"/>
Primitive Permutation Groups library.  In particular, a group
was wrongly labelled as PGL(2, 49).
</Item>

<Item>
There was a possible error in <C>SubgroupsSolvableGroup</C> when computing 
subgroups within a subgroup.
</Item>

<Item>
An error in 2-Cohomology computation for pc groups was fixed.
</Item>

<Item>
<C>IsConjugate</C> used normality in a wrong supergroup
</Item>
</List>


Fixed bugs which could lead to crashes:

<List>

<Item>
&GAP; crashed when the 
<C>PATH</C> environment variable was not set. [Reported by Robert F. Morse]
</Item>

<Item>
&GAP; could crash when started with option <C>-x 1</C>. 
Now the number of columns is initialized with at
least&nbsp;2. [Reported by Robert F. Morse]
</Item>

<Item>
After loading a saved workspace &GAP; 
crashed when one tried to slice
a compressed vector over a field with 2 &lt; q &lt;= 256 elements, which had
already existed in the saved workspace. [Reported by Laurent Bartholdi]
</Item>

<Item>
<C>FFECONWAY.WriteOverSmallestCommonField</C> tripped up when the common 
field is smaller than the field over which some of the vector elements 
are written, because it did a test based on the degree of the element, 
not the field it is written over. [Reported by Thomas Breuer]
</Item>

<Item>

Fixed the following error:
When an FFE in the Conway polynomial representation actually lied in a 
field that is handled in the internal representation (eg <M>GF(3)</M>) 
and you tried to write it over a bigger field that is ALSO handled 
internally (eg <M>GF(9)</M>) you got an element written over the larger 
field, but in the Conway polynomial representation, which is forbidden.
 [Reported by Jack Schmidt]
</Item>

<Item>
Attempting to compress a vector containing elements of a small finite field
represented as elements of a bigger (external) field caused a
segfault. [Reported by Edmund Robertson]
</Item>

<Item>
&GAP; crashed when <C>BlistList</C> was called with a range and a list 
containing large integers or non-integers. [Reported by Laurent Bartholdi]
</Item>

<Item>
&GAP; no longer crashes when <C>OnTuples</C> is called with a list that 
contains holes. [Reported by Thomas Breuer]
</Item>
</List>


Other fixed bugs:

<List>

<Item>
<C>Socle</C> for the trivial group could produce an error message.
</Item>

<Item>
<Ref BookName="ref" Func="DirectoryContents"/> ran into an error for immutable strings
without trailing slash as argument. [Reported by Thomas Breuer]
</Item>

<Item>
The functions <Ref BookName="ref" Func="IsInjective"/>
and <Ref BookName="ref" Func="IsSingleValued"/>
did not work for general linear mappings with trivial (pre)image. 
[Reported by Alper Odabas]
</Item>

<Item>
Creating an enumerator for a prime field with more than 65536 elements ran 
into an infinite recursion. [Reported by Akos Seress]
</Item>

<Item>
The performance of <C>List</C>, <C>Filtered</C>, <C>Number</C>, 
<C>ForAll</C> and <C>ForAny</C> if
applied to non-internally represented lists was improved. Also 
the performance of iterators for lists was slightly improved.
</Item>

<Item>
Finite field elements now know that they can be sorted easily which
improves performance in certain lookups.
</Item>

<Item>
A method for <Ref BookName="ref" Func="IsSubset"/>
was missing for the case that exactly one argument
is an inhomogeneous list. [Reported by Laurent Bartholdi]
</Item>

<Item>
Long integers in expressions are now printed (was not yet implemented). 
[Reported by Thomas Breuer]
</Item>

<Item>
Fixed kernel function for printing records.
</Item>

<Item>
New C library interfaces (e.g., to ncurses in the <Package>Browse</Package>
package) need some more memory to be allocated with <C>malloc</C>. The 
default value of &GAP; <C>-a</C> option is now <C>2m></C>.
</Item>

<Item>
Avoid warnings about pointer types by newer gcc compilers.
</Item>

<Item>
<C>IsBound(l[pos])</C> was failing for a large integer <A>pos</A> 
only when coded (e.g. in a loop or function body).
</Item>

<Item>
<C>ZmodpZObj</C> is now a synonym for
<C>ZmodnZObj</C> such that from now on such
objects print in a way that can be read back into &GAP;.
</Item>

<Item>
The outdated note that binary streams are not yet implemented has been
removed.
</Item>
</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% -->

<Section Label="fix4411">
<Heading>&GAP; 4.4 Update 11 (December 2008)</Heading>

Fixed bugs which could produce wrong results:

<List>

<Item>
<Ref BookName="ref" Func="MemoryUsage"/> on objects with no subobjects left them in the cache and
thus reported 0 in subsequent calls to MemoryUsage for the same
object. [Reported by Stefan Kohl]
</Item>

<Item>
<Ref BookName="ref" Func="Irr"/> might be missing characters. [Reported by Angel del Rio]
</Item>

<Item>
Up to now, it was allowed to call the function
<Ref BookName="ref" Func="FullMatrixAlgebraCentralizer"/>
with a field and a list of matrices such that the entries of the matrices
were not contained in the field;
in this situation, the result did not fit to the documentation.
Now the entries of the matrices are required to lie in the field,
if not then an error is signaled.
</Item>

<Item>
For those finite fields that are regarded as field extensions over
non-prime fields (one can construct such fields with
<Ref BookName="ref" Func="AsField"/>),
the function <Ref BookName="ref" Func="DefiningPolynomial"/>
erroneously returned a polynomial w.r.t. the extension of 
the prime field. [Reported by Stefan Kohl]
</Item>

<Item>
Since the release of &GAP; 4.4.10,
the return values of the function
<Ref BookName="ref" Func="QuaternionAlgebra"/>
were not consistent w.r.t. the attribute
<Ref BookName="ref" Func="GeneratorsOfAlgebra"/>;
the returned list could have length four or five.
Now always the list of elements of the canonical basis is returned.
</Item>

<Item>
<Ref BookName="ref" Func="MonomialGrevlexOrdering"/>
calculated a wrong ordering in certain cases. [Reported by Paul Smith]
</Item>

<Item>
The (&GAP; kernel) method for the operation
<Ref BookName="ref" Func="IntersectSet"/> for ranges had two bugs,
which could yield a result range with either too few or too many elements.
As a consequence, for example the <Ref BookName="ref" Func="Intersection"/>
results for ranges could be
wrong. [Reported by Matthew Fayers]
</Item>

<Item>
Fixed a bug in the short-form display of elements of larger finite fields, a bug in some
cross-field conversions and some inefficiencies and a missing method in the 
<Ref BookName="ref" Func="LogFFE"/> code. [Reported by Jia Huang]
</Item>

<Item>
In rare cases <Ref BookName="ref" Func="SmithNormalFormIntegerMatTransforms"/>
returned a wrong normal form (the version without transforming matrices 
did not have this problem). This is fixed. [Reported by Alexander Hulpke]
</Item>

<Item>
The variant of the function <Ref BookName="ref" Func="StraightLineProgram"/>
that takes a string
as its first argument returned wrong results if the last character of this
string was a closing bracket.
</Item>

<Item>
The code for central series in a permutation group used too tight a bound
and thus falsely return a nilpotent permutation group as non-nilpotent.
</Item>
</List>


Fixed bugs which could lead to crashes:

<List>

<Item>
Under certain circumstances the kernel code for position in blists
would access a memory location just after the end of the blist. If
this location was not accessible, a crash could result. This was corrected
and the code was cleaned up. [Reported by Alexander Hulpke]
</Item>
</List>


Other fixed bugs:

<List>

<Item>
The function <Ref BookName="ref" Func="IsomorphismTypeInfoFiniteSimpleGroup"/>
can be called with a positive integer instead of a group,
and then returns information about the simple group(s) of this order.
(This feature is currently undocumented.)
For the argument 1, however, it ran into an infinite loop.
</Item>

<Item>
A lookup in an empty dictionary entered a break loop. 
Now returns <C>fail</C>. [Reported by Laurent Bartholdi]
</Item>

<Item>
The c++ keyword <C>and</C> can no longer be used as a macro parameter 
in the kernel. [Reported by Paul Smith]
</Item>

<Item>
The operation <Ref BookName="ref" Func="KernelOfMultiplicativeGeneralMapping"/> has
methods designed to handle maps between permutation groups in a two-step approach,
but did not reliably trigger the second step.  This has now been fixed, preventing
a slow infinite loop repeating the first step.  This was normally only seen
as part of a larger calculation.
</Item>

<Item>
There were two methods for the operation
<Ref BookName="ref" Func="Intersection2"/>
which have implicitly assumed
that finiteness of a collection can always be decided. Now, these methods
check for <Ref BookName="ref" Func="IsFinite"/> and
<Ref BookName="ref" Func="CanComputeSize"/> prior to calling
<Ref BookName="ref" Func="IsFinite"/>.
</Item>

<Item>
Made error message in case of corrupted help book information
(manual.six file) shorter and more informative.  [Reported by Alexander Hulpke]
</Item>

<Item>
&GAP; cannot call methods with more than six
arguments.
Now the functions <Ref BookName="ref" Func="NewOperation"/>,
<Ref BookName="ref" Func="DeclareOperation"/>, and
<Ref BookName="ref" Func="InstallMethod"/>
signal an error if one attempts to declare an operation or to install
a method with more than six arguments.
</Item>

<Item>
Up to now, <Ref BookName="ref" Func="IsOne"/>
had a special method for general
mappings, which was much worse than the generic method;
this special method has now been removed.
</Item>

<Item>
When printing elements of an algebraic extension parentheses around
coefficients were missing. [Reported by Maxim Hendriks]
</Item>
</List>


New or improved functionality:

<List>

<Item>
Make dynamic loading of modules possible on CYGWIN using a DLL based
approach. Also move to using autoconf version 2.61.
</Item>

<Item>
One can now call
<Ref BookName="ref" Func="Basis"/>,
<Ref BookName="ref" Func="Iterator"/>
etc. with the return value of the function
<Ref BookName="ref" Func="AlgebraicExtension"/>.
</Item>

<Item>
The function <Ref BookName="ref" Func="FrobeniusCharacterValue"/>
returned <C>fail</C> for results that require
a finite field with more than 65536 elements.
Meanwhile &GAP; can handle larger finite fields,
so this restriction was removed.
(It is still possible that
<Ref BookName="ref" Func="FrobeniusCharacterValue"/> returns <C>fail</C>.)
</Item>

<Item>
Methods for testing membership in general linear groups and special linear
groups over the integers have been added. 
</Item>

<Item>
Methods for <Ref BookName="ref" Func="String"/> and
<C>ViewString</C> for full row modules have
been added. Further, a default method for
<Ref BookName="ref" Func="IsRowModule"/>
has been
added, which returns 
<C>false</C> for objects
which are not free left modules.
</Item>

<Item>
A <C>ViewString</C> method for objects with name
has been added.
</Item>

<Item>
The method for <Ref BookName="ref" Func="View"/>
for polynomial rings has been improved, and methods for
<Ref BookName="ref" Func="String"/> and
<C>ViewString</C> for polynomial rings
have been added.
</Item>

<Item>
<Ref BookName="ref" Func="Binomial"/> now works with huge <C>n</C>.
</Item>

<Item>
The function <Ref BookName="ref" Func="InducedClassFunctionsByFusionMap"/>
is now documented.
</Item>

<Item>
The return values of the function
<Ref BookName="ref" Func="QuaternionAlgebra"/>
now store that they are division rings (if optional parameters are given
then of course ths depends on these parameters).
</Item>
</List>

</Section>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% -->

<Section Label="fix4412">
<Heading>&GAP; 4.4 Update 12 (December 2008)</Heading>

Fixed bugs which could lead to crashes:

<List>
<Item>
A bug whereby leaving an incomplete statement on a 
line (for instance typing while and then return)
when prompt colouring was in use could lead to &GAP;
crashing.
</Item>
</List>

Other fixed bugs:
<List>
<Item>
A bug which made the command-line editor unusable
in a 64-bit version of &GAP; on Mac&nbsp;OS&nbsp;X.
</Item>
</List>

</Section>

</Chapter>

<!-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% --> 
<!-- %% --> 
<!-- %E -->