<Chapter Label="ChangesGAP46toGAP47"> 
<Heading>Changes between &GAP; 4.6 and &GAP; 4.7</Heading>

This chapter contains an overview of most important changes 
introduced in &GAP; 4.7.2 release (the first public release
of &GAP; 4.7).

It also contains information about subsequent update releases for &GAP; 4.7.

<Section Label="fix472"> 
<Heading>&GAP; 4.7.2 (December 2013)</Heading>

<Subsection Label="Changes in the core GAP47 system"> 
<Heading>Changes in the core &GAP; system introduced in &GAP; 4.7</Heading>

Improved and extended functionality:
<List>
<!-- cf. #411, 412, 413 -->
<Item>
The methods for computing conjugacy classes of permutation groups have 
been rewritten from scratch to enable potential use for groups in 
different representations. As a byproduct the resulting code is 
(sometimes notably) faster. It also now is possible to calculate 
canonical conjugacy class representatives in permutation groups, 
which can be beneficial when calculating character tables.
</Item>
<Item>
The methods for determining (conjugacy classes of) subgroups in 
non-solvable groups have been substantially improved in speed and 
scope for groups with multiple nonabelian composition factors. 
</Item>
<Item>
There is a new method for calculating the maximal subgroups of 
a permutation group (with chief factors of width less or equal 5) 
without calculating the whole subgroup lattice.
</Item>
<Item>
If available, information from the table of marks library is 
used to speed up subgroup calculations in almost simple factor groups.
</Item>
<Item>
The broader availability of maximal subgroups is used to improve 
the calculation of double cosets.
</Item>
<Item>
To illustrate the improvements listed above, one could try, for example
<Log><![CDATA[
g:=WreathProduct(MathieuGroup(11),Group((1,2)));
Length(ConjugacyClassesSubgroups(g));
]]></Log>
and
<Log><![CDATA[
g:=SemidirectProduct(GL(3,5),GF(5)^3);
g:=Image(IsomorphismPermGroup(g));
MaximalSubgroupClassReps(g);
]]></Log>
</Item>
<Item>
Computing the exponent of a finite group <M>G</M> could be extremely slow.
This was due to a slow default method being used, which computed
all conjugacy classes of elements in order to compute the exponent.
We now instead compute Sylow subgroups <M>P_1</M>, ..., <M>P_k</M> of 
<M>G</M> and use the easily verified equality 
<M>exp(G) = exp(P_1) x ... x exp(P_k)</M>. 
This is usually at least as fast and in many cases orders of magnitude 
faster. 
<Log><![CDATA[
gap> G:=SmallGroup(2^7*9,33);;
gap> H:=DirectProduct(G, ElementaryAbelianGroup(2^10));;
gap> Exponent(H); # should take at most a few milliseconds
72
gap> K := PerfectGroup(2688,3);;
gap> Exponent(K); # should take at most a few seconds
168
]]></Log>
</Item>
<Item>
<!-- #414 -->
The functionality in &GAP; for transformations and transformation semigroups has
been rewritten and extended. Partial permutations and inverse semigroups have
been newly implemented. The documentation for transformations and
transformation semigroups has been improved. Transformations and partial
permutations are implemented in the &GAP; kernel. Methods for calculating
attributes of transformations and partial permutations, and taking products,
and so are also implemented in the kernel. The new implementations are largely
backwards compatible; some exceptions are given below. 
<P/>
The degree of a transformation <C>f</C> is usually defined as the
largest positive integer where <C>f</C> is defined.  In previous versions of
&GAP;, transformations were only defined on positive integers less than their
degree, it was only possible to multiply transformations of equal degree, and a
transformation did not act on any point exceeding its degree.  Starting with
&GAP; 4.7, transformations behave more like permutations, in that
they fix unspecified points and it is possible to multiply arbitrary
transformations.
<List>
<Item>
in the display of a transformation, the trailing fixed points are no longer
printed. More precisely, in the display of a transformation <C>f</C> if
<C>n</C> is the largest value such that <C>n^f&lt;>n</C> or
<C>i^f=n</C> for some <C>i&lt;>n</C>, then the values exceeding <C>n</C>
are not printed.
</Item>
<Item>
the display for semigroups of transformations now includes more information,
for example <C>&lt;transformation semigroup on 10 pts with 10 generators></C> and 
<C>&lt;inverse partial perm semigroup on 10 pts with 10 generators></C>.
</Item>
<Item>
transformations which define a permutation can be inverted, and groups of
transformations can be created. 
</Item>
</List>
Further information regarding transformations and partial permutations, can be
found in the relevant chapters of the reference manual. 
<P/>
The code for Rees matrix semigroups has been completely rewritten to fix the
numerous bugs in the previous versions. The display of a Rees matrix semigroup
has also been improved to include the numbers of rows and columns, and the
underlying semigroup.  Again the new implementations should be backwards
compatible with the exception that the display is different.  
<P/>
The code for magmas with a zero adjoined has been improved so that it is
possible to access more information about the original magma. The display has
also been changed to indicate that the created magma is a magma with zero
adjoined (incorporating the display of the underlying magma). Elements of a
magma with zero are also printed so that it is clear that they belong to a
magma with zero. 
<P/>
If a semigroup is created by generators in the category
IsMultiplicativeElementWithOneCollection and CanEasilyCompareElements, then it
is now checked if the One of the generators is given as a generator. In this
case, the semigroup is created as a monoid.  
</Item>
<Item>
Added a new operation <Ref Oper="GrowthFunctionOfGroup" BookName="ref"/>
that gives sizes of distance spheres in the Cayley graph of a group.
</Item>
<Item>
<!-- #123 -->
A new group constructor <Ref Oper="FreeAbelianGroup" BookName="ref"/>
for free abelian groups has been added. By default, it creates suitable
fp groups. Though free abelian groups groups do not offer much
functionality right now, in the future other implementations may be
provided, e.g. by the <Package>Polycyclic</Package> package.
</Item>
<Item>
<!-- #29 -->
The message about halving the pool size at startup is only shown
when <C>-D</C> command line option is used 
(see <Ref Sect="Command line options" BookName="ref"/>).
[Suggested by Volker Braun]
</Item>
<Item>
<!-- #180 -->
An info class called <Ref InfoClass="InfoObsolete" BookName="ref"/> 
with the default level 0 is introduced.
Setting it to 1 will trigger warnings at runtime if an obsolete variable
declared with <C>DeclareObsoleteSynonym</C> is used. This is recommended for 
testing &GAP; distribution and packages.
</Item>
<Item>
The &GAP; help system now recognises some common different spelling patterns
(for example, -ise/-ize, -isation/-ization, solvable/soluble) and searches
for all possible spelling options even when the synonyms are not declared.
</Item>
<Item>
<!-- #452 -->
Added new function <Ref Func="Cite" BookName="ref"/> which produces citation 
samples for &GAP; and packages.
</Item>
<Item>
<!-- #198 -->
It is now possible to compile &GAP; with user-supplied <C>CFLAGS</C>
which now will not be overwritten by &GAP; default settings. 
[Suggested by Jeroen Demeyer]
</Item>
</List>

Fixed bugs:

<List>
<Item>
<!-- #444 -->
<Ref Func="Union" BookName="ref"/> had <M>O(n^3)</M> behaviour when 
given many ranges (e.g. it could take 10 seconds to find a union of 
1000 1-element sets). The new implementation reduces that to 
<M>O(n log n)</M> (and 4ms for the 10 second example), at the cost 
of not merging ranges as well as before in some rare cases.
</Item>
<Item>
<!-- #423 -->
<C>IsLatticeOrderBinaryRelation</C> only checked the existence of upper 
bounds but not the uniqueness of the least upper bound (and dually for 
lower bounds), so in some cases it could return the wrong answer.
[Reported by Attila Egri-Nagy]
</Item>
<Item>
<!-- #453 -->
<Ref Oper="LowIndexSubgroupsFpGroup" BookName="ref"/> triggered a
break loop if the list of generators of the 2nd argument contained 
the identity element of the group.
[Reported by Ignat Soroko]
</Item>
<Item>
<!-- #454 -->
Fixed regression in heuristics used by 
<Ref Oper="NaturalHomomorphismByNormalSubgroup" BookName="ref"/> that
could produce a permutation representation of an unreasonably large degree.
[Reported by Izumi Miyamoto]
</Item>
<Item>
<!-- #129 -->
Fixed inconsistent behaviour of <C>QuotientMod( Integers, r, s, m )</C>
in the case where <A>s</A> and <A>m</A> are not coprime. This fix also 
corrects the division behaviour of <C>ZmodnZ</C> objects, see 
<Ref Oper="QuotientMod" BookName="ref" /> and 
<Ref Func="ZmodnZ" BookName="ref"/>.
[Reported by Mark Dickinson]
</Item>
<Item>
<!-- #408 -->
Fixed an oversight in the loading process causing 
<Ref Func="OnQuit" BookName="ref"/> not resetting the options 
stack after exiting the break loop.
</Item>
<Item>
<!-- #407 -->
Empty strings were treated slightly differently than other strings in
the &GAP; kernel, for historical reasons. This resulted in various
inconsistencies. For example, <C>IsStringRep("")</C> returned true, 
but a method installed for arguments of type <C>IsStringRep</C> would 
NOT be invoked when called with an empty string. 
<P/>
We remove this special case in the &GAP; kernel (which dates back the 
very early days of &GAP;&nbsp;4 in 1996). This uncovered one issue in 
the kernel function <C>POSITION_SUBSTRING</C> (when calling it with an 
empty string as second argument), which was also fixed.
</Item>
<Item>
<!-- #219 -->
The parser for floating point numbers contained a bug that could cause 
&GAP; to crash or to get into a state where the only action left to the 
user was to exit &GAP; via Ctrl-D. For example, entering four dots with 
spaces between them on the &GAP; prompt and then pressing the return key 
caused &GAP; to exit.
<P/>
The reason was (ironically) an error check in the innards of the float
parser code which invoked the &GAP; <C>Error()</C> function at a point 
where it should not have.
</Item>
<Item>
Removing the last character in a string was supposed to overwrite the old
removed character in memory with a zero byte, but failed to do so due to an
off-by-one error. For most &GAP; operations, this has no visible effect, 
except for those which directly operate on the underlying memory representation 
of strings. For example, when trying to use such a string to reference a record
entry, a (strange) error could be triggered.
</Item>
<Item>
<Ref Oper="ViewString" BookName="ref"/> and <Ref Oper="DisplayString" BookName="ref"/>
are now handling strings, characters and immediate FFEs in a consistent manner.
</Item>
<Item>
<!-- #417, 436, 437, 446  -->
Multiple fixes to the build process for less common Debian platforms 
(arm, ia64, mips, sparc, GNU/Hurd). [Suggested by Bill Allombert]
</Item>
<Item>
<!-- #425, 426, 427, 438 -->
Fixes for several regressions in the <F>gac</F> script.
[Suggested by Bill Allombert]
</Item>
</List>

Changed functionality:
<List>  
<Item>
<!-- #230 -->
It is not possible now to call <Ref Oper="WreathProduct" BookName="ref"/> 
with 2nd argument <A>H</A> not being a permutation group, without using the 3rd 
argument specifying the permutation representation. This is an incompatible 
change but it will produce an error instead of a wrong result. The former
behaviour of <Ref Oper="WreathProduct" BookName="ref"/> may now be achieved
by using <Ref Oper="StandardWreathProduct" BookName="ref"/> which returns the wreath 
product for the (right regular) permutation action of <A>H</A> on its elements.
</Item>
<Item>
<!-- #180 -->
The function <C>ViewLength</C> to specify the maximal number of lines
that are printed in <Ref Oper="ViewObj" BookName="ref" /> became 
obsolete, since there was already a user preference <C>ViewLength</C> 
to specify this. The value of this preference is also accessible in
<C>GAPInfo.ViewLength</C>.
</Item>
</List>

</Subsection>

<Subsection Label="New and updated packages since GAP 4.6.5"> 
<Heading>New and updated packages since &GAP; 4.6.5</Heading>

<Index>Packages, new</Index>
At the time of the release of &GAP; 4.6.5 there were 107 packages 
redistributed with &GAP;. The first public release of &GAP;&nbsp;4.7 
contains 114 packages.
<P/>

One of essential changes is that
the <Package>Citrus</Package> package by J.Mitchell has been renamed 
to <Package>Semigroups</Package>. The package has been completely 
overhauled, the performance has been improved, and the code has been 
generalized so that in the future the same code can be used to compute 
with other types of semigroups.
<P/>

Furthermore, new packages that have been added to the redistribution 
since the release of &GAP; 4.6.5 are: 

<List>
<Item>
<Package>4ti2interface</Package> package by Sebastian Gutsche, providing
an interface to <Package>4ti2</Package>, a software package for algebraic, 
geometric and combinatorial problems on linear spaces (<URL>http://www.4ti2.de</URL>).
</Item>
<Item>
<Package>CoReLG</Package> by Heiko Dietrich, Paolo Faccin and 
Willem de Graaf for calculations in real semisimple Lie algebras.
</Item>
<Item>
<Package>IntPic</Package> package by Manuel Delgado, aimed at providing
a simple way of getting a pictorial view of sets of integers. The main 
goal of the package is producing <Package>Tikz</Package> code for arrays 
of integers. The code produced is to be included in a &LaTeX; file, which 
can then be processed. Some of the integers are emphasized by using 
different colors for the cells containing them.
</Item>
<Item>
<Package>LieRing</Package> by Serena Cicalo and Willem de Graaf for
constructing finitely-presented Lie rings and calculating the Lazard 
correspondence. The package also provides a database of small 
<M>n</M>-Engel Lie rings. 
</Item>
<Item>
<Package>LiePRing</Package> package by Michael Vaughan-Lee and Bettina 
Eick, introducing a new datastructure for nilpotent Lie rings of prime-power 
order. This allows to define such Lie rings for specific primes as well 
as for symbolic primes and other symbolic parameters. The package also 
includes a database of nilpotent Lie rings of order at most <M>p^7</M> 
for all primes <M>p > 3</M>.
</Item>
<Item>
<Package>ModIsom</Package> by Bettina Eick, which contains various
methods for computing with nilpotent associative algebras. In particular,
it contains a method to determine the automorphism group and to test
isomorphisms of such algebras over finite fields and of modular group
algebras of finite <M>p</M>-groups. Further, it contains a nilpotent quotient
algorithm for finitely presented associative algebras and a method
to determine Kurosh algebras.
</Item>
<Item>
<Package>SLA</Package> by Willem de Graaf for computations with 
simple Lie algebras. The main topics of the package are nilpotent 
orbits, theta-groups and semisimple subalgebras.
</Item>
</List>

<Index>Packages, upgraded</Index>
Furthermore, some packages have been upgraded substantially 
since the &GAP;&nbsp;4.6.5 release:
<List>
<Item>
<Package>ANUPQ</Package> package by Greg Gamble, Werner Nickel 
and Eamonn O'Brien has been updated after Max Horn joined it as a 
maintainer. As a result, it is now much easier to install and use
it with the current &GAP; release.
</Item>
<Item>
<Package>Wedderga</Package> package by Osnel Broche Cristo, Allen Herman, 
Alexander Konovalov, Aurora Olivieri, Gabriela Olteanu, Ángel del Río and 
Inneke Van Gelder has been extended to include functions for calculating
local and global Schur indices of ordinary irreducible characters of 
finite groups, cyclotomic algebras over abelian number fields, and rational
quaternion algebras (contribution by Allen Herman).
</Item>
</List>

</Subsection>

</Section>

<Section Label="fix473"> 
<Heading>&GAP; 4.7.3 (February 2014)</Heading>

Fixed bugs which could lead to incorrect results:
<List>
<Item>
Incorrect result returned by <C>AutomorphismGroup(PSp(4,2^n))</C>.
[Reported by Anvita]
</Item>
<Item>
The <Ref Oper="Order" BookName="ref"/> method for group homomorphisms 
newly introduced in &GAP;&nbsp;4.7 had a bug that caused it to 
sometimes return incorrect results.
[Reported by Benjamin Sambale]
</Item>
</List>

Fixed bugs that could lead to break loops:
<List>
<Item>
Several bugs were fixed and missing methods were introduced in the 
new code for transformations, partial permutations and semigroups 
that was first included in &GAP;&nbsp;4.7. Some minor corrections 
were made in the documentation for transformations.
</Item>
<Item>
Break loop in <C>IsomorphismFpMonoid</C> when prefixes in
generators names were longer than one letter.
[Reported by Dmytro Savchuk and Yevgen Muntyan]
</Item>
<Item>
Break loop while displaying the result of 
<Ref Func="MagmaWithInversesByMultiplicationTable" BookName="ref"/>.
[Reported by Grahame Erskine]
</Item>
</List>

Improved functionality:
<List>
<Item>
Better detection of UTF-8 terminal encoding on some systems.
[Suggested by Andries Brouwer]
</Item>
</List>

</Section>

<Section Label="fix474">
<Heading>&GAP; 4.7.4 (February 2014)</Heading>
This release was prepared immediately after &GAP; 4.7.3 to revert
the fix of the error handling for the single quote at the end of
an input line, contained in &GAP; 4.7.3. It happened that (only
on Windows) the fix caused error messages in one of the packages.
</Section>

<Section Label="fix475"> 
<Heading>&GAP; 4.7.5 (May 2014)</Heading>

Fixed bugs which could lead to incorrect results:
<List>
<Item>
<Ref Func="InstallValue" BookName="ref" /> cannot handle immediate 
values, characters or booleans for technical reasons. A check for 
such values was introduced to trigger an error message and prevent
incorrect results caused by this. [Reported by Sebastian Gutsche]
</Item>
<Item>
<Ref Oper="KnowsDictionary" BookName="ref"/> and 
<Ref Oper="LookupDictionary" BookName="ref" /> 
methods for <C>IsListLookupDictionary</C> were using 
<Ref Oper="PositionFirstComponent" BookName="ref" />;
the latter is only valid on sorted lists, but in 
<C>IsListLookupDictionary</C> the underlying 
list is NOT sorted in general, leading to bogus results.
</Item>
</List>

Other fixed bugs:
<List>
<Item>
A bug in <C>DirectProductElementsFamily</C> which used 
<Ref Prop="CanEasilyCompareElements" BookName="ref"/> instead of
<Ref Prop="CanEasilySortElements" BookName="ref"/>.
</Item>
<Item>
Fixed wrong <C>Infolevel</C> message that caused a break loop
for some automorphism group computations.
</Item>
<Item>
Fixed an error that sometimes caused a break loop in 
<Ref Func="HallSubgroup" BookName="ref"/>.
[Reported by Benjamin Sambale]
</Item>
<Item>
Fixed a rare error in computation of conjugacy classes of a
finite group by homomorphic images, providing fallback to a 
default algorithm.
</Item>
<Item>
Fixed an error in the calculation of Frattini subgroup in the
case of the trivial radical.
</Item>
<Item>
Several minor bugs were fixed in the documentation, kernel, and
library code for transformations.
</Item>
<Item>
Fixed errors in <Ref Func="NumberPerfectGroups" BookName="ref" />
and <Ref Func="NumberPerfectLibraryGroups" BookName="ref" /> not
being aware that there are no perfect groups of odd order.
</Item>
<Item>
Restored the ability to build &GAP; on OS X 10.4 and 10.5 
which was accidentally broken in the previous &GAP; release by
using the build option not supported by these versions.
</Item>
<Item>
Fixed some problems for ia64 and sparc architectures.
[Reported by Bill Allombert and Volker Braun]
</Item>
</List>

New package added for the redistribution with &GAP;:
<List>
<Item>
<Package>permut</Package> package by A.Ballester-Bolinches, 
E.Cosme-Llópez, and R.Esteban-Romero to deal with permutability 
in finite groups.
</Item>
</List>

</Section>


<Section Label="fix476"> 
<Heading>&GAP; 4.7.6 (November 2014)</Heading>

Fixed bugs which could lead to incorrect results:
<List>
<Item>
A bug that may cause <Ref Func="ShortestVectors" BookName="ref"/> to
return an incomplete list. [Reported by Florian Beye]
</Item>
<Item>
A bug that may lead to incorrect results and infinite loops when &GAP;
is compiled without GMP support using gcc 4.9.
</Item>
<Item>
A bug that may cause <Ref Func="OrthogonalEmbeddings" BookName="ref"/> to
return an incomplete result. [Reported by Benjamin Sambale]
</Item>
</List>

Fixed bugs that could lead to break loops:
<List>
<Item>
<Ref Oper="ClosureGroup" BookName="ref"/> should be used instead of
<Ref Func="ClosureSubgroup" BookName="ref"/> in case there is no parent
group, otherwise some calculations such as e.g. 
<Ref Oper="NormalSubgroups" BookName="ref"/> may fail.
[Reported by Dmitrii Pasechnik]
</Item>
<Item>
Fixed a line in the code that used a hard-coded identity permutation,
not a generic identity element of a group. [Reported by Toshio Sumi]
</Item>
<Item>
Fixed a problem in the new code for calculating maximal subgroups
that caused a break loop for some groups from the transitive groups
library. [Reported by Petr Savicky]
</Item>
<Item>
Fixed a problem in <Ref Oper="ClosureSubgroup" BookName="ref"/> not
accepting some groups without <Ref Oper="Parent" BookName="ref"/>.
[Reported by Inneke van Gelder]
</Item>
</List>

Other fixed bugs:
<List>
<Item>
Eliminated a number of compiler warnings detected with some newer 
versions of <Package>C</Package> compilers.
</Item>
<Item>
Some minor bugs in the transformation and partial permutation code 
and documentation were resolved. 
</Item>
</List>

</Section>


<Section Label="fix477"> 
<Heading>&GAP; 4.7.7 (February 2015)</Heading>

New features:
<List>
<Item>
Introduced some arithmetic operations for infinity and negative infinity,
see <Ref Label="Infinity" BookName="ref"/>.
</Item>
<Item>
Introduced new property <Ref Prop="IsGeneratorsOfSemigroup" BookName="ref"/>
which reflects wheter the list or collection generates a semigroup.
</Item>
</List>

Fixed bugs which could lead to incorrect results:
<List>
<Item>
Fixed a bug in <Ref Func="Union" BookName="ref"/> (actually,
in the internal library function <C>JoinRanges</C>) caused by 
downward running ranges. [Reported by Matt Fayers]
</Item>
<Item>
Fixed a bug where recursive records might be printed with the wrong 
component name, coming from component names being ordered differently
in two different pieces of code. [Reported by Thomas Breuer]
</Item>
<Item> 
The usage of <C>abs</C> in <File>src/gmpints.c</File> was replaced by 
<C>AbsInt</C>. The former is defined to operate on 32-bit integers
even if &GAP; is compiled in 64-bit mode. That lead to truncating
&GAP; integers and caused a crash in <Ref Func="RemInt" BookName="ref"/>, 
reported by Willem De Graaf and Heiko Dietrich. Using <C>AbsInt</C> fixes 
the crash, and ensures the correct behaviour on 32-bit and 64-bit builds.
</Item>
</List>

Fixed bugs that could lead to break loops:
<List>
<Item>
A problem with <Ref Func="ProbabilityShapes" BookName="ref"/> not setting
frequencies list for small degrees. [Reported by Daniel Błażewicz and
independently by Mathieu Gagne]
</Item>
<Item>
An error when generating a free monoid of rank infinity. 
[Reported by Nick Loughlin]
</Item>
<Item>
Several bugs with the code for Rees matrix semigroups not handling
trivial cases properly. 
</Item>
<Item>
A bug in <Ref Attr="IsomorphismTypeInfoFiniteSimpleGroup" BookName="ref"/>
affecting one particular group due to a misformatting in a routine that 
translates between the Chevalley type and the name used in the table 
(in this case, <C>"T"</C> was used instead of <C>["T"]</C>).
[Reported by Petr Savicky]
</Item>
</List>

Other fixed bugs:
<List>
<Item>
The <Ref Oper="Basis" BookName="ref"/> method for full homomorphism
spaces of linear mappings did not set basis vectors which could be
obtained by <Ref Attr="GeneratorsOfLeftModule" BookName="ref"/>.
</Item>
<Item>
A problem with <Ref Attr="GaloisType" BookName="ref"/> entering an
infinite loop in the routine for approximating a root. [Reported by Daniel Błażewicz]
</Item>
<Item>
Fixed the crash when &GAP; is called when the environment variables
<C>HOME</C> or <C>PATH</C> are unset. [Reported by Bill Allombert]
</Item>
</List>

Furthermore, new packages that have been added to the redistribution 
since the release of &GAP; 4.7.6 are: 

<List>
<Item>
<Package>json</Package> package by Christopher Jefferson, providing
a mapping between the <Package>JSON</Package> markup language and &GAP;
</Item>
<Item>
<Package>SglPPow</Package> package by Bettina Eick and Michael Vaughan-Lee, 
providing the database of <M>p</M>-groups of order <M>p^7</M> for 
<M>p &gt; 11</M>, and of order <M>3^8</M>.
</Item>
</List>

</Section>


<Section Label="fix478"> 
<Heading>&GAP; 4.7.8 (June 2015)</Heading>

</Section>

Fixed bugs which could lead to incorrect results:
<List>
<Item>
Added two groups of degree 1575 which were missing in the 
library of first primitive groups. [Reported by Gordon Royle]
</Item>
<Item>
Fixed the error in the code for algebra module elements in packed 
representation caused by the use of <Ref Oper="Objectify" BookName="ref"/>
with the type of the given object instead of <Ref Func="ObjByExtRep" BookName="ref"/>
as recommended in <Ref Sect="Further Improvements in Implementing Residue Class Rings"
BookName="ref"/>. The problem was that after calculating <C>u+v</C> where one of the
summands was known to be zero, this knowledge was wrongly passed to the sum via the 
type. [Reported by Istvan Szollosi]
</Item>
<Item>
Fixed a bug in <Ref Oper="PowerMod" BookName="ref"/> causing wrong results 
for univariate Laurent polynomials when the two polynomial arguments are 
stored with the same non-zero shift. [Reported by Max Horn]
</Item>
</List>

Furthermore, new packages that have been added to the redistribution 
since the release of &GAP; 4.7.7 are: 

<List>
<Item>
<Package>PatternClass</Package> by Michael Albert, Ruth Hoffmann and 
Steve Linton, allowing to explore the permutation pattern classes 
build by token passing networks. Amongst other things, it can compute 
the basis of a permutation pattern class, create automata from token 
passing networks and check if the deterministic automaton is a possible 
representative of a token passing network.
</Item>
<Item>
<Package>QPA</Package> by Edward Green and Øyvind Solberg, providing
data structures and algorithms for computations with finite dimensional 
quotients of path algebras, and with finitely generated modules over such 
algebras. It implements data structures for quivers, quotients of path 
algebras, and modules, homomorphisms and complexes of modules over 
quotients of path algebras.
</Item>
</List>

</Chapter>


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