(new-style) packages for saclib 2.2.6 and qepcad-B.1.69

[sagetrac-bgoodri]

There exists a new upstream versions of saclib and qepcad that compile on 64 bit machines at https://www.usna.edu/CS/qepcadweb/INSTALL/IQ.html

Sources:

• http://www.usna.edu/CS/qepcadweb/INSTALL/saclib2.2.6.tar.gz
• http://www.usna.edu/CS/qepcadweb/INSTALL/qepcad-B.1.69.tar.gz

Copy both tarballs into your $SAGE_ROOT/upstream directory and rename saclib2.2.6.tar.gz to saclib-2.2.6.tar.gz

There could be some improvements to the QEPCAD interface, but it would probably be better for me to open separate tickets and base them on the updated spkg.

Another thing that would be good to include is Chris Brown's SLFQ program

https://www.usna.edu/CS/qepcadweb/SLFQ/Home.html

which calls QEPCAD B. I have gotten SLFQ to build outside of sage with a couple of patches, but I have not actually gotten it to work yet.

CC: @jasongrout @sagetrac-cwitty @jondo

Component: packages: optional

Keywords: qepcad

Author: Thierry Monteil

Branch: 4c26982

Reviewer: Jeroen Demeyer, Matthias Koeppe, Dima Pasechnik

Issue created by migration from https://trac.sagemath.org/ticket/10224

[sagetrac-maldun]

comment:2

Works fine for me. (Ubuntu 10.10)

But did you make a Mercurial update of your changes?
If I build your package I get:

Created package qepcad-1.54.p0.spkg.

NAME: qepcad

VERSION: 1.54.p0
SIZE: 4.0M
HG REPO: Unchecked in changes
SPKG.txt: Good

Btw: Version 1.55 is out, perhaps try to use the new source.

Regards,
Stefan

[kcrisman]

comment:3

If this ever wants to leave experimental, it will need some changes. See for instance this sage-support thread, where it seems some of the install scripts are tcsh (?!?), and #11933 confirms it doesn't build everywhere. The spkg-install will need work, too - there is no checking whether saclib even builds correctly before going on!

This spkg link does still work for now, though, which is good!

[jondo]

comment:7

Updated the links in the description.

[fchapoton]

Changed keywords from none to qepcad

[kcrisman]

comment:10

Note that #16642 and https://groups.google.com/forum/#!topic/sage-support/X6U8FxnNuzU continue to confirm this optional package does not compile everywhere...

[jondo]

comment:11

I again updated the links to the QEPCAD pages.

[mkoeppe]

comment:12

There's a fork of qepcad with a CMake-based build system here: https://github.com/PetterS/qepcad

Still does not compile on Mac OS X because of gnu-libc-isms such as using "ieee754.h"

(sage-sh) mkoeppe@d245:qepcad$ cmake .
-- The C compiler identification is GNU 4.9.2
-- The CXX compiler identification is GNU 4.9.2
-- Checking whether C compiler has -isysroot
-- Checking whether C compiler has -isysroot - yes
-- Checking whether C compiler supports OSX deployment target flag
-- Checking whether C compiler supports OSX deployment target flag - yes
-- Check for working C compiler: /Users/mkoeppe/s/sage/sage-6.5/local/bin/gcc
-- Check for working C compiler: /Users/mkoeppe/s/sage/sage-6.5/local/bin/gcc -- works
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Detecting C compile features
-- Detecting C compile features - done
-- Checking whether CXX compiler has -isysroot
-- Checking whether CXX compiler has -isysroot - yes
-- Checking whether CXX compiler supports OSX deployment target flag
-- Checking whether CXX compiler supports OSX deployment target flag - yes
-- Check for working CXX compiler: /Users/mkoeppe/s/sage/sage-6.5/local/bin/g++
-- Check for working CXX compiler: /Users/mkoeppe/s/sage/sage-6.5/local/bin/g++ -- works
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Detecting CXX compile features
-- Detecting CXX compile features - done
-- No build type specified; defaulting to CMAKE_BUILD_TYPE=Release.
-- Configuring done
-- Generating done
-- Build files have been written to: /Users/mkoeppe/s/qepcad
(sage-sh) mkoeppe@d245:qepcad$ make
Scanning dependencies of target saclib
[  0%] Building C object CMakeFiles/saclib.dir/saclib/src/AADV.c.o
In file included from /Users/mkoeppe/s/qepcad/saclib/include/saclib.h:11:0,
                 from /Users/mkoeppe/s/qepcad/saclib/src/AADV.c:13:
/Users/mkoeppe/s/qepcad/saclib/include/hfloats.h:23:21: fatal error: ieee754.h: No such file or directory
 #include <ieee754.h>
                     ^
compilation terminated.
make[2]: *** [CMakeFiles/saclib.dir/saclib/src/AADV.c.o] Error 1
make[1]: *** [CMakeFiles/saclib.dir/all] Error 2
make: *** [all] Error 2

[jdemeyer]

comment:36

Replying to @sagetrac-tmonteil:

I propose to add a "not tested" flag to those examples in order not to pollute the tutorial, and then add tests that transforms the string to a set of substrings so that we can test equality. Does it makes sense ?

Better use # random instead of # not tested if you want to run the tests but not check the output.

[jdemeyer]

comment:37

Also reviewer name is missing...

[dimpase]

comment:38

OK, I'd be happy to give a positive review once the change proposed by Jeroen is done.

[sagetrac-tmonteil]

comment:39

Replying to @jdemeyer:

Better use # random instead of # not tested if you want to run the tests but not check the output.

Actually, the tests tagged # not tested are run in the TESTS: section that follows the EXAMPLE: section where it appears, but there, the output is sent to a function that makes a set of atoms so that the difference of ordering (that seems arch dependent) is taken into account. So, be both have meaningful user-friendly examples and the doctest are still executed and partially checked. Hence, if i replace # not tested with # random in the EXAMPLE: section, those tests will be run twice and take longer. I do not mind doing this, but i do not see the point since any test is already run at least once. Tell me.

[jdemeyer]

comment:40

OK, I see your point.

One reason to still have # random is to make it more explicit that the output of the test is random. When a test says "not tested", it is not clear why it is not tested. At the very least, write something like # not tested (random order).

How much time do those tests take? If we're talking about a few seconds, then I would just use # random.

And the reviewer name is still missing.

[dimpase]

Reviewer: Jeroen Demeyer, Matthias Koeppe, Dima Pasechnik

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

e0b1caf

#10224 : make explicit why some examples are not tested (random order).

[sagetrac-git]

Changed commit from 0535b3a to e0b1caf

[sagetrac-tmonteil]

comment:43

Replying to @jdemeyer:

At the very least, write something like # not tested (random order).

I like this alternative, both explicit and not redundant.

[novoselt]

comment:44

Replying to @dimpase:

OK, I'd be happy to give a positive review once the change proposed by Jeroen is done.

So - it is done and can be positive reviewed? (I don't really follow what is going on here, but I will add these packages to SageMathCell when they are ready.)

[dimpase]

comment:45

trying this with the latest beta, and seeing that one needs to set "make -j1" in order to build qepcad successfully. At least for me...

[dimpase]

comment:46

added one more commit to indicate the latter issue in SPKG.txt

[dimpase]

Changed branch from u/tmonteil/saclib_2_2_6_qepcad_B_1_69 to public/10224

[dimpase]

Changed commit from e0b1caf to 4c26982

[vbraun]

Changed branch from public/10224 to 4c26982

[novoselt]

comment:48

Given that it fails consistently with "make -j2" and most likely parallel building is enabled, shouldn't the build script enforce "make -j1" rather than mention it somewhere?..

[novoselt]

Changed commit from 4c26982 to none

[rwst]

comment:49

Replying to @novoselt:

Given that it fails consistently with "make -j2" and most likely parallel building is enabled, shouldn't the build script enforce "make -j1" rather than mention it somewhere?..

Exactly, please review #19450.

[jondo]

comment:50

This issue's component is packages: optional. Currently the package is 'experimental'.
In order to promote it to 'optional', it needs to build on all supported systems. Is this already the case?