[sage-devel] Blocker tickets needs review
Hello, this is a reminder that currently 3 blocker tickets need review. All of them have some discussion, but they all seem stalled: * Set JUPYTER_CONFIG_DIR https://trac.sagemath.org/ticket/21430 * Old installed version of Cython is used https://trac.sagemath.org/ticket/21441 * CoinBackend: _test_solve fails on 32-bit https://trac.sagemath.org/ticket/21449 There is a fourth blocker ticket with ongoing work: * No emacs interface anymore in sage 7.4 beta0: both sage_mode and sage-shell-mode broken https://trac.sagemath.org/ticket/21227 -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at https://groups.google.com/group/sage-devel. For more options, visit https://groups.google.com/d/optout.
Re: [sage-devel] Blocker tickets
On 04/24/2013 04:52 PM, Martin Raum wrote: Implementing a workaround for all OS's then is essentially equivalent to skipping part of the test I don't agree with this statement. The test is about the functioning of pexpect, not about OS-specific timings. It's not the job of Sage to test the operating system, it's the job of Sage to test Sage components. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sage-devel] Blocker tickets
I must say that I saw the ticket and was about to review it, because nobody had done this so far. But I read Volker's comment, and I agree with him. Perhaps the reason for the failure of all of us to review this ticket is that not only Volker feels uncomfortable with such a change. I don't know of any bad consequences of the approach taken, either. But here is the reason why I still disagree with it. On a functioning OS, we would expect the workaround to be unnecessary. Implementing a workaround for all OS's then is essentially equivalent to skipping part of the test, namely correct and timely shutdown of Sage. So we would sacrifice parts of test without need to. I also agree with Volker's argument on maintainability. On the other hand, I am sure it won't be the end of the world, if the ticket gets merged as is. I just wouldn't want to give a positive review to a ticket that I personally disagree with. Am Mittwoch, 24. April 2013 14:45:32 UTC+2 schrieb Jeroen Demeyer: > > On 04/24/2013 02:37 PM, Volker Braun wrote: > > For the record, I disagree with the approach to both #14371 and #14460 > > in that fugly workarounds to fundamental bugs in Solaris or a particular > > gcc version are applied liberally to, basically, everyone. Sure, > > sometimes it is necessary to work around bugs, but that should be a > > special case and not the generic code path. > > Do you know of actual bad consequences of the approach taken at #14371 > or #14460? I would not apply ugly workarounds to everyone *if* they have > bad consequences. If those workarounds don't really influence existing > working setups (which is the case with the mentioned patches), then > what's the problem? In this case, it's better to be safe and apply the > workaround on all systems. > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sage-devel] Blocker tickets
I don't think there is a bad consequence for repeatedly checking the status of a terminated child process with exponential backoff. Its just really weird that you would ever do that if your OS works correctly. And IMHO it hurts code maintainability. Not to mention debugging, assuming that you are on a sane platform when you are stepping through that. I agree that this is more of a philosophical objection than something that directly affects functionality. If somebody else is happy with that then fine. It just doesn't feel right for me to put my stamp of approval on it. On Wednesday, April 24, 2013 1:45:32 PM UTC+1, Jeroen Demeyer wrote: > > On 04/24/2013 02:37 PM, Volker Braun wrote: > > For the record, I disagree with the approach to both #14371 and #14460 > > in that fugly workarounds to fundamental bugs in Solaris or a particular > > gcc version are applied liberally to, basically, everyone. Sure, > > sometimes it is necessary to work around bugs, but that should be a > > special case and not the generic code path. > > Do you know of actual bad consequences of the approach taken at #14371 > or #14460? I would not apply ugly workarounds to everyone *if* they have > bad consequences. If those workarounds don't really influence existing > working setups (which is the case with the mentioned patches), then > what's the problem? In this case, it's better to be safe and apply the > workaround on all systems. > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sage-devel] Blocker tickets
On 04/24/2013 02:37 PM, Volker Braun wrote: For the record, I disagree with the approach to both #14371 and #14460 in that fugly workarounds to fundamental bugs in Solaris or a particular gcc version are applied liberally to, basically, everyone. Sure, sometimes it is necessary to work around bugs, but that should be a special case and not the generic code path. Do you know of actual bad consequences of the approach taken at #14371 or #14460? I would not apply ugly workarounds to everyone *if* they have bad consequences. If those workarounds don't really influence existing working setups (which is the case with the mentioned patches), then what's the problem? In this case, it's better to be safe and apply the workaround on all systems. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sage-devel] Blocker tickets
For the record, I disagree with the approach to both #14371 and #14460 in that fugly workarounds to fundamental bugs in Solaris or a particular gcc version are applied liberally to, basically, everyone. Sure, sometimes it is necessary to work around bugs, but that should be a special case and not the generic code path. On Wednesday, April 24, 2013 1:18:13 PM UTC+1, Jeroen Demeyer wrote: > > *ping* > > This sage-5.9 blocker still needs review: > > #14371: Race condition in singular doctest > http://trac.sagemath.org/sage_trac/ticket/14371 > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sage-devel] Blocker tickets
*ping* This sage-5.9 blocker still needs review: #14371: Race condition in singular doctest http://trac.sagemath.org/sage_trac/ticket/14371 -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sage-devel] Blocker tickets
This still needs review: #14371: Race condition in singular doctest http://trac.sagemath.org/sage_trac/ticket/14371 And also the GCC-4.8.0 work-around would be good to merge in sage-5.9 (but not a blocker): http://trac.sagemath.org/sage_trac/ticket/14460 -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sage-devel] Blocker tickets
Hi, Tickt 14433 fails on my system: sage:len(search_doc('tree', interact=False).splitlines() 3956 sage: version() 'Sage Version 5.8, Release Date: 2013-03-15' bg, Johannes On 12.04.2013 16:01, Jeroen Demeyer wrote: > There are 2 blocker tickets remaining for Sage 5.9, both of them need > review: > > #14371: Race condition in singular doctest > http://trac.sagemath.org/sage_trac/ticket/14371 > > #14426: Runaway/Segfaulting ECL processes > http://trac.sagemath.org/sage_trac/ticket/14426 > > > There is already 1 blocker for Sage 5.10, which also needs review: > #14433: Increase number of matches to "tree" in sagedoc.py doctest > http://trac.sagemath.org/sage_trac/ticket/14433 > -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
[sage-devel] Blocker tickets
There are 2 blocker tickets remaining for Sage 5.9, both of them need review: #14371: Race condition in singular doctest http://trac.sagemath.org/sage_trac/ticket/14371 #14426: Runaway/Segfaulting ECL processes http://trac.sagemath.org/sage_trac/ticket/14426 There is already 1 blocker for Sage 5.10, which also needs review: #14433: Increase number of matches to "tree" in sagedoc.py doctest http://trac.sagemath.org/sage_trac/ticket/14433 -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
Re: [sage-devel] Blocker tickets needing review
On 05/ 4/11 08:50 AM, Jeroen Demeyer wrote: On 2011-05-03 21:44, Volker Braun wrote: I agree that sage-4.7 should compile with gcc-4.6.[01]. But why not specifically excluding these two instead of an wildcard? What if gcc 4.6.2 exhibits the same bug? I think the *safer* option is to assume the status-quo that the bug will not be fixed. If it does get fixed, we can still change the spkg. I am cc'ed on the relevant gcc tickets, so I will know when the bugs get fixed. Jeroen. I agree with Jeroen. I'd much rather see the fix included until such time as we know the bug has been removed from gcc, rather than hope some distant version fixes the bug. I've known gcc bugs to go on a for a long time - especaily those only affecting rarer platforms. -- A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail? -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Blocker tickets needing review
On 2011-05-03 21:44, Volker Braun wrote: > I agree that sage-4.7 should compile with gcc-4.6.[01]. But why not > specifically excluding these two instead of an wildcard? I did this for cliquer because that gcc bug should be fixed, see #11227. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Blocker tickets needing review
The *safest* option of them all is of course to never build with optimizations. Its comparably easy to generate straightforward machine code, the compiler bugs are invariably tied to the optimization process. Also, I think that old bugs in the gcc bugzilla are not systematically tested against new releases. Yes the bug might be closed if a developer works on that specific bug, but it might just get solved as a sideeffect of general infrastructure improvements. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Blocker tickets needing review
On 2011-05-03 21:44, Volker Braun wrote: > I agree that sage-4.7 should compile with gcc-4.6.[01]. But why not > specifically excluding these two instead of an wildcard? What if gcc 4.6.2 exhibits the same bug? I think the *safer* option is to assume the status-quo that the bug will not be fixed. If it does get fixed, we can still change the spkg. I am cc'ed on the relevant gcc tickets, so I will know when the bugs get fixed. Jeroen. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Blocker tickets needing review
I agree that sage-4.7 should compile with gcc-4.6.[01]. But why not specifically excluding these two instead of an wildcard? -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Blocker tickets needing review
On 2011-05-03 17:23, Volker Braun wrote: > I have a bad feeling about this. We both know that nobody is going to > check whether its still necessary to downgrade optimizations by the time > that, say, 4.6.5 rolls out. Those bugs have been reported upstream to gcc, so if they get fixed, I will get notified and I can change the spkg accordingly. > IHMO it would be much better to use the > compiler wrapper and limit optimizations globally for specific compiler > releases. Probably, but the compiler wrapper is not in sage-4.7 and I would still like to have sage-4.7 compile with gcc 4.6.0 and a possible future gcc 4.6.1. I think it is better to have a workaround for a fixed bug than to have no workaround for an existing bug. Jeroen. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Blocker tickets needing review
I have a bad feeling about this. We both know that nobody is going to check whether its still necessary to downgrade optimizations by the time that, say, 4.6.5 rolls out. IHMO it would be much better to use the compiler wrapper and limit optimizations globally for specific compiler releases. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Blocker tickets needing review
On 2011-05-03 15:43, Volker Braun wrote: > Wait just because there may be a bug in 4.6.0 we disable optimizations > for future gcc versions that may fix these? If these future gcc versions are released, we can change the spkgs accordingly. At least a pre-release version of gcc 4.6.1 still contains the bugs (as was pointed out on sage-release). Jeroen. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Blocker tickets needing review
Wait just because there may be a bug in 4.6.0 we disable optimizations for future gcc versions that may fix these? -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
Re: [sage-devel] Blocker tickets needing review
On 2011-05-03 14:34, Jeroen Demeyer wrote: > There are two blockers for sage-4.7 which need review, both are spkg > patches: > > #11226: Sympow spkg fails with gcc 4.6.0 > > #11278: singular 3-1-1-4.p8 fails on Mac OS X 10.4 Two more blocker tickets needing review, where a check for gcc version 4.6.0 was replaced by a check for gcc version 4.6.* #11168: rubiks fails doctest with gcc 4.6.0 and -O2 optimisation. #11227: gcc-4.6.0 optimization bug with cliquer-1.2.p7.spkg -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
[sage-devel] Blocker tickets needing review
There are two blockers for sage-4.7 which need review, both are spkg patches: #11226: Sympow spkg fails with gcc 4.6.0 #11278: singular 3-1-1-4.p8 fails on Mac OS X 10.4 Normally, these two tickets are the last obstacles for a sage-4.7 release. Jeroen. -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org