Karma swaps
I have two updates for F37 that will not be stable by final freeze
based on time alone, but that I think should be stable to avoid
disruptions to users. I could wait and ask for freeze exceptions, but
I would first like to ask the community to give karma to these
updates. I am willing to test and give karma to your updates, too.
Let me know what updates to test, and how to test them.
The two updates are:
https://bodhi.fedoraproject.org/updates/FEDORA-2022-a6d8261b07 - sympy
https://bodhi.fedoraproject.org/updates/FEDORA-2022-21f6ca9264 - gap,
pari, & friends
The sympy update deals with the disappearance of the JDK on i386.
Sympy uses the antlr4 package, a Java package, to generate some
grammar files. Since there is no JDK on i386, there is no antlr4
package either. This makes one of the BuildRequires for the package
unresolvable. The ideal solution would be to stop building sympy on
i386. However, sympy is consumed by a number of packages that still
build for i386. I think that is the right long-term goal, and will
work toward that in the coming months. This update is a stop-gap
measure. It switches sympy from a noarch package to an archful
package so that we can omit the problematic BuildRequires on i386.
This will leave sympy partially broken on i386 (it is still mostly
functional), but that is unavoidable at this point. I think this
should go stable so that the noarch -> archful change happens at a
release boundary, rather than after F37 is released. (And hopefully
we can change back for F38!)
For testing purposes, this build of sympy should function identically
to the previous build. Install the python3-sympy and python3-ipython
packages, then verify that it starts and runs at all, perhaps like
this:
```
$ isympy
[snip sympy banner]
In [1]: import math
In [2]: math.sqrt(9)
Out[2]: 3.0
In [3]: from sympy import symbols
In [4]: x, y = symbols('x y')
In [5]: expr = x + 2*y
In [6]: expr
Out[6]: x + 2⋅y
```
The second update is large, containing 100 package builds. It
contains an update of gap to version 4.12.0, and an update of pari to
version 2.15.0. Both new versions come with significant performance
improvements. The gap update in particular drops i386 builds for all
packages named gap-pkg-*, as well as gap, GAPDoc, and xgap. I think
this should go stable before release so that the i386 disappearing act
doesn't happen post-release, and also because the gap and pari updates
both contain a few backwards incompatibilities, which should be
introduced at release boundaries.
Almost every package in this update has a test suite that is run at
build time, so they have been tested quite well already, and I have
been testing them also. To test the gap part, install the gap and
gap-pkg-xmod packages. (The gap-pkg-xmod package pulls in several
other gap-pkg-* packages, so we can show they all work together.)
Then do a few things with gap-pkg-xmod, like this:
```
$ gap
[snip gap banner]
gap> LoadPackage("xmod");
[snip package loading messages]
gap> c5 := Group( (5,6,7,8,9) );
Group([ (5,6,7,8,9) ])
gap> SetName( c5, "c5" );
gap> id5 := IdentityMapping( c5 );
IdentityMapping( c5 )
gap> ac5 := AutomorphismGroup( c5 );
<group with 1 generator>
gap> act := MappingToOne( c5, ac5 );
[ (5,6,7,8,9) ] -> [ IdentityMapping( c5 ) ]
gap> XMod( id5, act ) = XModByBoundaryAndAction( id5, act );
true
gap> H12 := SmallCatOneGroup( 12, 4, 3 );
Cat-1-group with underlying group Group( [ f1, f2, f3 ] ) .
gap> C12 := SmallCat1Group( 12, 4, 3 );
[Group( [ f1, f2, f3 ] ) => Group( [ f1, f2, <identity> of ... ] )]
gap> C12 := CatOneGroupToXMod( H12 );
[Group( [ f1, f2, f3 ] ) => Group( [ f1, f2, <identity> of ... ] )]
gap> C18 := Cat1Select( 18, 4, 3 );
[(C3 x C3) : C2 => Group( [ f1, <identity> of ..., <identity> of ... ] )]
gap> H18 := Cat1GroupToHAP( C18 );
Cat-1-group with underlying group (C3 x C3) : C2 .
gap> IdCatOneGroup( H18 );
[ 18, 4, 3 ]
gap> IdCat1Group( C18 );
[ 18, 4, 3 ]
```
To test the pari part, install the pari-gp package and try a few calculations,
like this:
```
$ gp
[snip pari banner]
? 1 + 1
%1 = 2
? (x+1)^(-2)
%2 = 1/(x^2 + 2*x + 1)
? Mod(x, x^2+x*y+y^2)^3
%3 = Mod(y^3, x^2 + y*x + y^2)
? T = x^2 + 1;
? factor(T*(1 + O(5^3)))
%5 =
[ (1 + O(5^3))*x + (2 + 5 + 2*5^2 + O(5^3)) 1]
[(1 + O(5^3))*x + (3 + 3*5 + 2*5^2 + O(5^3)) 1]
? R(x) = {
my(s);
s = zeta(2) * sum(a=1, sqrt(x), moebius(a)*(x\a^2));
(s - x) / x^0.4;
}
? R(10^7)
%7 = 0.052092560787004188970344062406837112269
? R(10^12)
%8 = 0.010948893958117048274619354741656931605
? R(10^15)
%9 = 0.00012562581350676279794919752938422775380
```
Warning: you'll notice a slight pause when computing R(10^12), and a
significant pause when computing R(10^15). Pari hasn't hung. Wait a
minute and it will be back with the answer.
Sorry for the epistle. I will be very grateful for any testing and
karma help you can give me. Let me know what I can test and give
karma to for you.
--
Jerry James
http://www.jamezone.org/