New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix Matrix_gfpn_dense * int #25076
Comments
comment:1
While we are at it: How is one supposed to run optional tests? It seems that "sage -t --optional=meataxe" will ONLY run the tests marked as optional, hence, it will skip all compulsory tests, and thus will fail in most tests as definitions given in the compulsory tests are missing when running the optional tests. |
comment:2
Replying to @simon-king-jena:
|
comment:4
Works for me:
|
Replying to @simon-king-jena:
Please read https://groups.google.com/forum/#!topic/sage-devel/H0t9l6XdfPI |
comment:6
Simon: can you specify which version of Sage you using and whether any patches are applied? |
comment:7
Replying to @jdemeyer:
Sorry. It's the branch from #18514, which means Sage 8.5.beta2. Another change: I recently installed tkinter. And something seems to be broken anyway: "git status" gives me untracked files After removing the misplaced directories, re-installing the packages and sage -br, I get
So, perhaps I should retry with #18514 rebased on top of the latest develop branch? |
comment:8
PS: The same error was observed by other people as well, on #18514. So, I have to see what went wrong there. |
comment:9
Replying to @simon-king-jena:
Yes and let me know if that fixes it. |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
Attachment: giac-1.4.9.45.p2.log Failing giac on Linux 4.4.0-116-generic #140-Ubuntu x86_64 |
comment:12
Replying to @jdemeyer:
Giac fails to build (see attached log). Known problem? |
comment:13
Replying to @simon-king-jena:
People have reported it but nobody knows how to reproduce it. |
comment:14
Replying to @jdemeyer:
Ouch. Any work-around (such as rebuilding from scratch)? Could it be related with tkinter being installed (which used to not be the case when I last built Sage)? |
comment:15
Replying to @simon-king-jena:
It's certainly something to try. Please let us know whether that fixes your giac problem. |
comment:16
I don't know why, but the two messages I was trying to send to sage-release via slrn didn't come through. Trying now make distclean; make on a clean develop branch. |
comment:17
OK, rebuilding from scratch did work. And with clean develop branch, I still get the error. So, what is special about matrix_gfpn_dense? What is special about multiplying it with an int? And why the heck is _mul_long not called (why has it been removed from sage.matrix in the first place)? |
This comment has been minimized.
This comment has been minimized.
comment:19
Now that's really frustrating. It is supposed to be related with coercion. Coercion is supposed to be machine independent. So, why (as Jeroen clarifies in the new ticket description) is the bug only present on some machines? By the way, on #18514, at least one other person has seen that error. So, it is not restricted to my own laptop. |
comment:20
If it makes you feel better, I managed to reproduce it on a different machine. |
comment:21
Replying to @simon-king-jena:
did you try other native python types, e.g. float? |
comment:22
it might be gmp vs mpir difference, by the way. |
comment:23
Replying to @dimpase:
Doesn't make sense, as a float doesn't coerce into GF(3). By inserting print statements into the bin_op method of the coercion model, I found that it isn't even called, although it should be. |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
comment:37
Replying to @simon-king-jena:
|
comment:38
Replying to @jdemeyer:
Issue 1 is fixed in the dependency and moreover these are not the only issues related with the original ticket description. And what would be the point of opening yet another ticket when the remaining issue can be dealt with here? It is, by the way, not without precedence that a ticket description was changed because of issues that came up while working on the original ticket description. |
comment:39
Replying to @simon-king-jena:
Because this ticket already does something useful. To turn your question around: what would be the point of not opening yet another ticket given that there are really two issues here (one about coercing scalars to the base ring and one about using |
Changed keywords from _mul_long to none |
This comment has been minimized.
This comment has been minimized.
Author: SimonKing |
comment:41
OK, just adding a test here, and then opening another ticket for the two remaining issues (that do belong together). New commits:
|
Commit: |
comment:42
Note that I constructed the test in such a way that blindly using the current implementation of _mul_long would give the wrong result, which means it also makes sure that the follow-up ticket will use the right conversion. |
comment:43
Replying to @simon-king-jena:
Those still look like different issues to me. One is about the coercion model and one is about the meataxe interface. |
comment:44
Replying to @simon-king-jena:
See #25079 |
comment:45
I think it would be good to refer to this ticket instead of #24742 and to add a test for |
comment:46
Replying to @jdemeyer:
Yes, they are two different issues, but it is totally natural to deal with them on a single ticket: When one only fixes the coercion model, one would actually introduce a bug (namely: The flawed _mul_long of Matrix_gfpn_dense would be used). When one only fixes the meataxe interface, one wouldn't be able to demonstrate that it is fixed without using cython in doctests. Yes, using cython in doctests is possible, but why should one, if the existing test "M*int(4)==M" is good enough? |
Branch pushed to git repo; I updated commit sha1. New commits:
|
Reviewer: Jeroen Demeyer |
Changed branch from u/SimonKing/fix_matrix_gfpn_dense___int to |
Changed commit from |
Changed author from SimonKing to Simon King |
The following previous bug was fixed by #24742:
This ticket adds a test that makes sure that the issue remains fixed.
Depends on #24742
CC: @tscrim
Component: coercion
Author: Simon King
Branch:
f0e97af
Reviewer: Jeroen Demeyer
Issue created by migration from https://trac.sagemath.org/ticket/25076
The text was updated successfully, but these errors were encountered: