-
-
Notifications
You must be signed in to change notification settings - Fork 517
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
libgap: distinct __str__ and __repr__ for GapElements #27946
Comments
comment:1
Here is a branch for this specific change that I extracted from the rest of my work on #18267, since this change is worth considering in its own right. Still needs a bit more testing to ensure it didn't inadvertently break any other tests. New commits:
|
Author: Erik Bray |
Commit: |
comment:2
Works without test failures for me. |
Reviewer: Vincent Delecroix |
Changed branch from u/embray/libs/gap/separate-str-repr to |
comment:5
Not in Sage 8.8. Let's please to try keep tickets' milestones related to the release in which we actually intend to include them, and in particular the release in which they were actually included, especially when closing tickets. |
### 📚 Description Trac branch `u/gh-collares/gap-gc` from #34701, now migrated to GitHub. Currently based atop #35093; will rebase once that is merged. The rest of the description below is copied from #34701: A refactor in #27946 introduced "unprotected" (not surrounded by `GAP_Enter`/`GAP_Leave`) `GAP_ValueGlobalVariable` calls. I believe this might be a GC hazard, because after updating to GAP 4.12.1 I started seeing aarch64 crashes on NixOS infrastructure such as: ``` #0 0x0000fffff79740e8 in wait4 () #1 0x0000fffff5dc6b78 in print_enhanced_backtrace () #2 0x0000fffff5dc8190 in sigdie () #3 0x0000fffff5dcb1c0 in cysigs_signal_handler () #4 0x0000fffff7ffb7cc in __kernel_rt_sigreturn () #5 0x0000ffff99a0bf28 in ConvString () #6 0x0000000000000000 in ?? () #7 0x0000000000000000 in ?? () #8 0x0000000000000000 in ?? () #9 0x0000ffff99989930 in Pr () #10 0x0000ffff9998aa18 in CloseOutput () #11 0x0000ffff99884828 in capture_stdout () at /build/sage- src-9.7/pkgs/sagemath-standard/sage/libs/gap/element.pyx:154 ... ``` I also see cases where `capture_stdout` throws errors such as `sage.libs.gap.util.GAPError: Error, Length: <list> must be a list (not the integer 255)` and then crashes. Both types of errors are fixed by this ticket. Note that I am nesting `GAP_Enter`/`GAP_Leave` calls because I didn't remove the preexisting calls inside `capture_stdout`. That's because I feared removing the innermost calls might create a new footgun (and I believe nested `GAP_Enter`/`GAP_Leave` calls are explicitly supported), but removing them should cause no problem. Removing them might even be preferable for performance reasons, I don't know. Fixes #34701 ### 📝 Checklist <!-- Put an `x` in all the boxes that apply. --> <!-- If your change requires a documentation PR, please link it appropriately --> <!-- If you're unsure about any of these, don't hesitate to ask. We're here to help! --> - [x] I have made sure that the title is self-explanatory and the description concisely explains the PR. - [x] I have linked an issue or discussion. - [ ] I have created tests covering the changes. - [ ] I have updated the documentation accordingly. ### ⌛ Dependencies <!-- List all open pull requests that this PR logically depends on --> <!-- - #xyz: short description why this is a dependency - #abc: ... --> - #35093: GAP 4.12.2 upgrade, which touches the same function and should land first. URL: #35114 Reported by: Mauricio Collares Reviewer(s): Dima Pasechnik
Currently the
libgap.GapElement
objects have a__repr__
(but no explicit__str__
, so the two are the same) which is taken from calling the GAP functionViewObj
on the object and capturing and returning the result. This is essentially the same as when you display the object in a GAP interpreter prompt without usingPrint()
, such as:Although GAP has many options for displaying/printing variables, this makes
ViewObj
the closest possible thing to the__repr__
method on Python objects, which is called in similar circumstances:This is in stark contrast with the pexpect interface's
GapElement
which (through a sequence of calls) effectively implements its__repr__
by callingPrint(obj)
. Arguably this was not the best choice, but changing it might be tricky at this point. However, it does not explicitly implement__str__
, soGapElement.__str__
does give a sensible result in this case.Arguably
Print(obj)
is the right choice for__str__
since again in rough comparison with Python, when youprint(...)
an object in Python its__str__
is called.This proposes to change just the libgap interface to meet this specification:
libgap.GapElement.__str__
captures the output ofPrint(obj)
and returns it. Whilelibgap.GapElement.__repr__
captures and returns the output ofViewObj(obj)
.This gives the best of both worlds, and better consistency in most contexts where GapElements are converted to strings, making replacing pexpect gap with libgap easier and less disruptive.
CC: @nthiery
Component: interfaces
Keywords: gap libgap
Author: Erik Bray
Branch/Commit:
63db722
Reviewer: Vincent Delecroix
Issue created by migration from https://trac.sagemath.org/ticket/27946
The text was updated successfully, but these errors were encountered: