Re: [sage-release] Re: Sage 8.8.rc0 released
On Wed, Jun 12, 2019 at 6:13 PM Volker Braun wrote: > > On Wednesday, June 12, 2019 at 2:05:04 PM UTC+2, E. Madison Bray wrote: >> >> One very easily solution to this, and I know some people will shudder >> (myself included) is to simply include the configure script in the >> repository. > > > We already have a way of bundling up the configure scripts in a tarball, we > may just as well use that. Its not used right now becaues a) monotonically > increasing version numbers for the confball are a pita, and b) we don't want > to put every temporary confball on the mirror network. E.g. if somebody steps > up and extends the bootstrap script to save/load based on the git head sha1, > e.g.: > > ./bootstrap --save-head > > saves upstream/configure-.tag.gz > > ./bootstrap --load-head http://path/to/upstream/ > > loads http://path/to/upstream/configure-.tag.gz > > then I can add them to the release script and buildbots... > OK, great, I've opened https://trac.sagemath.org/ticket/27978 for this. > -- > You received this message because you are subscribed to the Google Groups > "sage-release" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sage-release+unsubscr...@googlegroups.com. > To post to this group, send email to sage-release@googlegroups.com. > Visit this group at https://groups.google.com/group/sage-release. > To view this discussion on the web visit > https://groups.google.com/d/msgid/sage-release/9e00eb09-b985-4634-90b6-718931463a54%40googlegroups.com. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "sage-release" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-release+unsubscr...@googlegroups.com. To post to this group, send email to sage-release@googlegroups.com. Visit this group at https://groups.google.com/group/sage-release. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/CAAWYfq0rihCrs8EudY7rjnCiE29i0CEJKbTgYpAFHHzQMoUYAw%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [sage-release] Re: Sage 8.8.rc0 released
On Wednesday, June 12, 2019 at 2:05:04 PM UTC+2, E. Madison Bray wrote: > > One very easily solution to this, and I know some people will shudder > (myself included) is to simply include the configure script in the > repository. We already have a way of bundling up the configure scripts in a tarball, we may just as well use that. Its not used right now becaues a) monotonically increasing version numbers for the confball are a pita, and b) we don't want to put every temporary confball on the mirror network. E.g. if somebody steps up and extends the bootstrap script to save/load based on the git head sha1, e.g.: ./bootstrap --save-head saves upstream/configure-.tag.gz ./bootstrap --load-head http://path/to/upstream/ loads http://path/to/upstream/configure-.tag.gz then I can add them to the release script and buildbots... -- You received this message because you are subscribed to the Google Groups "sage-release" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-release+unsubscr...@googlegroups.com. To post to this group, send email to sage-release@googlegroups.com. Visit this group at https://groups.google.com/group/sage-release. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/9e00eb09-b985-4634-90b6-718931463a54%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
[sage-release] Re: Sage 8.8.rc0 released
On Ubuntu 18.04 running on bi-Xeon E5-2623 (8 cores) + 16 GB RAM: *- python2: * - build (-j16) from a fresh git clone + pull develop completed, with Sphinx complaining during the doc build: formatargspec() is now deprecated. Please use sphinx.util.inspect.Signature instead - all tests from ptestlong passed - *python3: * -build (-j16) from a fresh git clone + pull develop completed, with the same Sphinx deprecation warning as with python 2. - all tests from ptest-python3 passed Eric. -- You received this message because you are subscribed to the Google Groups "sage-release" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-release+unsubscr...@googlegroups.com. To post to this group, send email to sage-release@googlegroups.com. Visit this group at https://groups.google.com/group/sage-release. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/d378f3ea-4eb8-4026-86e0-fe5a457a7fe3%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
Re: [sage-release] Re: Sage 8.8.rc0 released
On Wed, Jun 12, 2019 at 2:42 PM E. Madison Bray wrote: > > On Wed, Jun 12, 2019 at 3:14 PM Dima Pasechnik wrote: > > > > On Wed, Jun 12, 2019 at 2:03 PM E. Madison Bray > > wrote: > > > > > > On Wed, Jun 12, 2019 at 2:46 PM Dima Pasechnik wrote: > > > > > > > > On Wed, Jun 12, 2019 at 1:05 PM E. Madison Bray > > > > wrote: > > > > > > > > > > On Wed, Jun 12, 2019 at 11:45 AM Dima Pasechnik > > > > > wrote: > > > > > > > > > > > > On Tue, Jun 11, 2019 at 7:52 AM E. Madison Bray > > > > > > wrote: > > > > > > > > > > > > > > I don't know why the rush to call this a release candidate when > > > > > > > there are clearly still several outstanding problems to be > > > > > > > resolved before the next release. > > > > > > > > > > > > > > As usual absolutely zero communication of a plan or coordination > > > > > > > with the community. > > > > > > > > > > > > > my ongoing complaint is the lack of progress on the improving the > > > > > > release procedures to make it easy to change the configure package. > > > > > > At the moment it is extremely sub-optimal, with endless merge > > > > > > conflicts resulting solely from the need to have the ticket branch > > > > > > that changes configure.ac, spkg-configure.m4 files, and its friends, > > > > > > such as sage/big/sage-env* things coming with its configure > > > > > > "package" > > > > > > tarball. This leaves tickets of this sort languishing for months, as > > > > > > it's not really possible to get them in sync with the ongoing > > > > > > release > > > > > > process (already due to configure tarballs coming with artificial > > > > > > version numbers, that should not clash etc). > > > > > > > > > > > > I continue to think that the release process should not require > > > > > > updates of configure package. Sage seems to unique in its insistence > > > > > > on the release process depending on the autogenerated dumps... > > > > > > > > > > One very easily solution to this, and I know some people will shudder > > > > > (myself included) is to simply include the configure script in the > > > > > repository. Yes, it's generated. But there's precedent for this. > > > > > > > > well, it would make running patchbots easier, but > > > > I don't see how this would help the release management, > > > > as auto-merging would be still be out of the question. > > > > > > > > The auto-generated crud is simply unmergable, by and large. > > > > > > That really depends on the change in question. Oftentimes they're > > > quite discrete. It only becomes a mess when you have to do lots of > > > AC_REQUIREs and thus things get reordered in the configure script. > > > > > > That is a problem for us right now since we're doing a lot of > > > heavy-lifting on configure. Once a lot of that is out of the way it > > > will calm down a bit. > > > > given the speed things at present move on #27330, it can easily take a > > year or two. > > And still, as we shift more into using external libraries, there is > > going to be more tweaking of > > the configure package than in the olden days... > > Absolutely; which is maybe a good reason to just include the configure > script in the repository. as well as all the stuff in config/ and build/make/Makefile-auto.in so it's more than just one file. > True, that would not fix all the merge > issues you're struggling with, but on some level that is almost > unavoidable, at least when testing/building on systems that are not > able to run autoconf smoothly. A more complex workaround involving a > separate build system just for autoconf is probably not worth the > effort compared to that one minor concession. it's a not a workaround - the current release process is a huge bottleneck for #27330 and it is simply buggy, as merging new/updated spkg-configure.m4 files and stuff in m4/ without updatng configure is error-prone. Having this in place would simplify things and make them more robust. If merging into WIP happens on a local box (I don't recall how Volker does it), and then the result is pushed to bots (or pulled by them) I really do not see a reason for it not being in place, it's totally trivial to implement. If the merging happens on github/lab, it is still not hard to put some commit hooks/TravisCI calls/whatever into place to make it happen. All this can be also made to create a configure tarball and push it somewhere. > > Meanwhile, keeping it in the repository will make it much easier to > maintain said "tweaks". > > I'm not going to fight tooth-and-nail for it, but I'll just emphasize > > 1) We'd be far from the first project to make this concession > 2) It will make enough peoples' lives easier that it's not so bad to > have one generated (plain text) file in the repo > > > > > For the most part, when making small tweaks to > > > a single autoconf macro, the resulting change in the generated > > > configure script is more-or-less a direct translation. > > > > > > So for fixing bugs in configure alone it would be
Re: [sage-release] Re: Sage 8.8.rc0 released
On Wed, Jun 12, 2019 at 3:14 PM Dima Pasechnik wrote: > > On Wed, Jun 12, 2019 at 2:03 PM E. Madison Bray wrote: > > > > On Wed, Jun 12, 2019 at 2:46 PM Dima Pasechnik wrote: > > > > > > On Wed, Jun 12, 2019 at 1:05 PM E. Madison Bray > > > wrote: > > > > > > > > On Wed, Jun 12, 2019 at 11:45 AM Dima Pasechnik > > > > wrote: > > > > > > > > > > On Tue, Jun 11, 2019 at 7:52 AM E. Madison Bray > > > > > wrote: > > > > > > > > > > > > I don't know why the rush to call this a release candidate when > > > > > > there are clearly still several outstanding problems to be resolved > > > > > > before the next release. > > > > > > > > > > > > As usual absolutely zero communication of a plan or coordination > > > > > > with the community. > > > > > > > > > > > my ongoing complaint is the lack of progress on the improving the > > > > > release procedures to make it easy to change the configure package. > > > > > At the moment it is extremely sub-optimal, with endless merge > > > > > conflicts resulting solely from the need to have the ticket branch > > > > > that changes configure.ac, spkg-configure.m4 files, and its friends, > > > > > such as sage/big/sage-env* things coming with its configure "package" > > > > > tarball. This leaves tickets of this sort languishing for months, as > > > > > it's not really possible to get them in sync with the ongoing release > > > > > process (already due to configure tarballs coming with artificial > > > > > version numbers, that should not clash etc). > > > > > > > > > > I continue to think that the release process should not require > > > > > updates of configure package. Sage seems to unique in its insistence > > > > > on the release process depending on the autogenerated dumps... > > > > > > > > One very easily solution to this, and I know some people will shudder > > > > (myself included) is to simply include the configure script in the > > > > repository. Yes, it's generated. But there's precedent for this. > > > > > > well, it would make running patchbots easier, but > > > I don't see how this would help the release management, > > > as auto-merging would be still be out of the question. > > > > > > The auto-generated crud is simply unmergable, by and large. > > > > That really depends on the change in question. Oftentimes they're > > quite discrete. It only becomes a mess when you have to do lots of > > AC_REQUIREs and thus things get reordered in the configure script. > > > > That is a problem for us right now since we're doing a lot of > > heavy-lifting on configure. Once a lot of that is out of the way it > > will calm down a bit. > > given the speed things at present move on #27330, it can easily take a > year or two. > And still, as we shift more into using external libraries, there is > going to be more tweaking of > the configure package than in the olden days... Absolutely; which is maybe a good reason to just include the configure script in the repository. True, that would not fix all the merge issues you're struggling with, but on some level that is almost unavoidable, at least when testing/building on systems that are not able to run autoconf smoothly. A more complex workaround involving a separate build system just for autoconf is probably not worth the effort compared to that one minor concession. Meanwhile, keeping it in the repository will make it much easier to maintain said "tweaks". I'm not going to fight tooth-and-nail for it, but I'll just emphasize 1) We'd be far from the first project to make this concession 2) It will make enough peoples' lives easier that it's not so bad to have one generated (plain text) file in the repo > > For the most part, when making small tweaks to > > a single autoconf macro, the resulting change in the generated > > configure script is more-or-less a direct translation. > > > > So for fixing bugs in configure alone it would be useful to keep the > > generated script in the repository and not have to mess with this > > configure tarball every time. > > > > -- > > You received this message because you are subscribed to the Google Groups > > "sage-release" group. > > To unsubscribe from this group and stop receiving emails from it, send an > > email to sage-release+unsubscr...@googlegroups.com. > > To post to this group, send email to sage-release@googlegroups.com. > > Visit this group at https://groups.google.com/group/sage-release. > > To view this discussion on the web visit > > https://groups.google.com/d/msgid/sage-release/CAOTD34b3SLks2hmA_RmF24ikyW3Z9JR-gi3LtnjuAnBurwfecQ%40mail.gmail.com. > > For more options, visit https://groups.google.com/d/optout. > > -- > You received this message because you are subscribed to the Google Groups > "sage-release" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sage-release+unsubscr...@googlegroups.com. > To post to this group, send email to sage-release@googlegroups.com. > Visit this group at
Re: [sage-release] Re: Sage 8.8.rc0 released
On Wed, Jun 12, 2019 at 2:03 PM E. Madison Bray wrote: > > On Wed, Jun 12, 2019 at 2:46 PM Dima Pasechnik wrote: > > > > On Wed, Jun 12, 2019 at 1:05 PM E. Madison Bray > > wrote: > > > > > > On Wed, Jun 12, 2019 at 11:45 AM Dima Pasechnik wrote: > > > > > > > > On Tue, Jun 11, 2019 at 7:52 AM E. Madison Bray > > > > wrote: > > > > > > > > > > I don't know why the rush to call this a release candidate when there > > > > > are clearly still several outstanding problems to be resolved before > > > > > the next release. > > > > > > > > > > As usual absolutely zero communication of a plan or coordination with > > > > > the community. > > > > > > > > > my ongoing complaint is the lack of progress on the improving the > > > > release procedures to make it easy to change the configure package. > > > > At the moment it is extremely sub-optimal, with endless merge > > > > conflicts resulting solely from the need to have the ticket branch > > > > that changes configure.ac, spkg-configure.m4 files, and its friends, > > > > such as sage/big/sage-env* things coming with its configure "package" > > > > tarball. This leaves tickets of this sort languishing for months, as > > > > it's not really possible to get them in sync with the ongoing release > > > > process (already due to configure tarballs coming with artificial > > > > version numbers, that should not clash etc). > > > > > > > > I continue to think that the release process should not require > > > > updates of configure package. Sage seems to unique in its insistence > > > > on the release process depending on the autogenerated dumps... > > > > > > One very easily solution to this, and I know some people will shudder > > > (myself included) is to simply include the configure script in the > > > repository. Yes, it's generated. But there's precedent for this. > > > > well, it would make running patchbots easier, but > > I don't see how this would help the release management, > > as auto-merging would be still be out of the question. > > > > The auto-generated crud is simply unmergable, by and large. > > That really depends on the change in question. Oftentimes they're > quite discrete. It only becomes a mess when you have to do lots of > AC_REQUIREs and thus things get reordered in the configure script. > > That is a problem for us right now since we're doing a lot of > heavy-lifting on configure. Once a lot of that is out of the way it > will calm down a bit. given the speed things at present move on #27330, it can easily take a year or two. And still, as we shift more into using external libraries, there is going to be more tweaking of the configure package than in the olden days... > For the most part, when making small tweaks to > a single autoconf macro, the resulting change in the generated > configure script is more-or-less a direct translation. > > So for fixing bugs in configure alone it would be useful to keep the > generated script in the repository and not have to mess with this > configure tarball every time. > > -- > You received this message because you are subscribed to the Google Groups > "sage-release" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sage-release+unsubscr...@googlegroups.com. > To post to this group, send email to sage-release@googlegroups.com. > Visit this group at https://groups.google.com/group/sage-release. > To view this discussion on the web visit > https://groups.google.com/d/msgid/sage-release/CAOTD34b3SLks2hmA_RmF24ikyW3Z9JR-gi3LtnjuAnBurwfecQ%40mail.gmail.com. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "sage-release" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-release+unsubscr...@googlegroups.com. To post to this group, send email to sage-release@googlegroups.com. Visit this group at https://groups.google.com/group/sage-release. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/CAAWYfq2AssVut_34C0d9i85mXkAFnJMiRRv7Q9QCoWKRmDNygg%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [sage-release] Re: Sage 8.8.rc0 released
On Wed, Jun 12, 2019 at 2:46 PM Dima Pasechnik wrote: > > On Wed, Jun 12, 2019 at 1:05 PM E. Madison Bray wrote: > > > > On Wed, Jun 12, 2019 at 11:45 AM Dima Pasechnik wrote: > > > > > > On Tue, Jun 11, 2019 at 7:52 AM E. Madison Bray > > > wrote: > > > > > > > > I don't know why the rush to call this a release candidate when there > > > > are clearly still several outstanding problems to be resolved before > > > > the next release. > > > > > > > > As usual absolutely zero communication of a plan or coordination with > > > > the community. > > > > > > > my ongoing complaint is the lack of progress on the improving the > > > release procedures to make it easy to change the configure package. > > > At the moment it is extremely sub-optimal, with endless merge > > > conflicts resulting solely from the need to have the ticket branch > > > that changes configure.ac, spkg-configure.m4 files, and its friends, > > > such as sage/big/sage-env* things coming with its configure "package" > > > tarball. This leaves tickets of this sort languishing for months, as > > > it's not really possible to get them in sync with the ongoing release > > > process (already due to configure tarballs coming with artificial > > > version numbers, that should not clash etc). > > > > > > I continue to think that the release process should not require > > > updates of configure package. Sage seems to unique in its insistence > > > on the release process depending on the autogenerated dumps... > > > > One very easily solution to this, and I know some people will shudder > > (myself included) is to simply include the configure script in the > > repository. Yes, it's generated. But there's precedent for this. > > well, it would make running patchbots easier, but > I don't see how this would help the release management, > as auto-merging would be still be out of the question. > > The auto-generated crud is simply unmergable, by and large. That really depends on the change in question. Oftentimes they're quite discrete. It only becomes a mess when you have to do lots of AC_REQUIREs and thus things get reordered in the configure script. That is a problem for us right now since we're doing a lot of heavy-lifting on configure. Once a lot of that is out of the way it will calm down a bit. For the most part, when making small tweaks to a single autoconf macro, the resulting change in the generated configure script is more-or-less a direct translation. So for fixing bugs in configure alone it would be useful to keep the generated script in the repository and not have to mess with this configure tarball every time. -- You received this message because you are subscribed to the Google Groups "sage-release" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-release+unsubscr...@googlegroups.com. To post to this group, send email to sage-release@googlegroups.com. Visit this group at https://groups.google.com/group/sage-release. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/CAOTD34b3SLks2hmA_RmF24ikyW3Z9JR-gi3LtnjuAnBurwfecQ%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
Re: [sage-release] Re: Sage 8.8.rc0 released
On Wed, Jun 12, 2019 at 1:46 PM Dima Pasechnik wrote: > > On Wed, Jun 12, 2019 at 1:05 PM E. Madison Bray wrote: > > > > On Wed, Jun 12, 2019 at 11:45 AM Dima Pasechnik wrote: > > > > > > On Tue, Jun 11, 2019 at 7:52 AM E. Madison Bray > > > wrote: > > > > > > > > I don't know why the rush to call this a release candidate when there > > > > are clearly still several outstanding problems to be resolved before > > > > the next release. > > > > > > > > As usual absolutely zero communication of a plan or coordination with > > > > the community. > > > > > > > my ongoing complaint is the lack of progress on the improving the > > > release procedures to make it easy to change the configure package. > > > At the moment it is extremely sub-optimal, with endless merge > > > conflicts resulting solely from the need to have the ticket branch > > > that changes configure.ac, spkg-configure.m4 files, and its friends, > > > such as sage/big/sage-env* things coming with its configure "package" > > > tarball. This leaves tickets of this sort languishing for months, as > > > it's not really possible to get them in sync with the ongoing release > > > process (already due to configure tarballs coming with artificial > > > version numbers, that should not clash etc). > > > > > > I continue to think that the release process should not require > > > updates of configure package. Sage seems to unique in its insistence > > > on the release process depending on the autogenerated dumps... > > > > One very easily solution to this, and I know some people will shudder > > (myself included) is to simply include the configure script in the > > repository. Yes, it's generated. But there's precedent for this. > > well, it would make running patchbots easier, but > I don't see how this would help the release management, > as auto-merging would be still be out of the question. > > The auto-generated crud is simply unmergable, by and large. > > The patchbots are largely an orthogonal issue here. ticket merging in the release process happens in batches, as Volker explained; half a dozen tickets are merged into a WIP branch and the result sent to the buildbots. The buildbots might not have autotools installed, so they cannot take a branch without a ready ./configure etc (i.e. a ready configure tarball, ATM) The merging of tickets with different configure tarballs, or even worse, with different ./configure scripts is, however, hard. What I don't understand is: why not plug into the process the post-processing of the merged WIP branch to actually run ./bootstrap, so that configure tarballs/ or ./configure & co made on the spot, before the stuff is sent to the buildbots to check. Where is the catch here, why the latter is not done? > > > > Among other projects, CPython includes their configure script in the > > repository. This allows developers to get right up-and-running with > > ./configure without having to fuss with the nuances of different > > versions of autotools on different systems. Only developers actually > > working on configure issues need to worry about that. It's quite > > convenient. > > > > > > > > On Tue, Jun 11, 2019, 02:28 Kwankyu Lee wrote: > > > >> > > > >> > > > >> > > > >> On Monday, June 10, 2019 at 4:40:52 PM UTC+9, Sébastien Labbé wrote: > > > >>> > > > >>> Testing the following optional and external packages: > > > >>> > > > >>> External software detected for doctesting: > > > >>> ffmpeg,graphviz,gurobi,imagemagick,internet,latex,pandoc > > > >>> > > > >>> Using > > > >>> --optional=4ti2,bliss,cbc,ccache,cmake,cryptominisat,dot2tex,e_antic,external,glucose,latte_int,lidia,lrslib,memlimit,mpir,ninja_build,normaliz,notedown,openssl,pandoc_attributes,pycosat,pynormaliz,python2,qhull,rst2ipynb,sage,topcom > > > >>> External software to be detected: > > > >>> cplex,ffmpeg,graphviz,gurobi,imagemagick,internet,latex,macaulay2,magma,maple,mathematica,matlab,octave,pandoc,scilab > > > >>> > > > >>> I get: > > > >>> > > > >>> -- > > > >>> sage -t --long src/sage/databases/findstat.py # 8 doctests failed > > > >>> sage -t --long src/sage/combinat/designs/ext_rep.py # 1 doctest > > > >>> failed > > > >>> -- > > > >>> > > > >>> and only the following is reproducible: > > > >>> > > > >>> -- > > > >>> sage -t --long src/sage/databases/findstat.py # 8 doctests failed > > > >>> -- > > > >>> > > > >>> Follow up at https://trac.sagemath.org/ticket/25536 > > > >> > > > >> > > > >> I see Sage has been much improved in this respect. Thank you for your > > > >> persistent efforts! > > > >> > > > >> -- > > > >> You received this message because you are subscribed to the Google > > > >> Groups "sage-release" group. > > > >> To unsubscribe from this
Re: [sage-release] Re: Sage 8.8.rc0 released
On Wed, Jun 12, 2019 at 1:05 PM E. Madison Bray wrote: > > On Wed, Jun 12, 2019 at 11:45 AM Dima Pasechnik wrote: > > > > On Tue, Jun 11, 2019 at 7:52 AM E. Madison Bray > > wrote: > > > > > > I don't know why the rush to call this a release candidate when there are > > > clearly still several outstanding problems to be resolved before the next > > > release. > > > > > > As usual absolutely zero communication of a plan or coordination with the > > > community. > > > > > my ongoing complaint is the lack of progress on the improving the > > release procedures to make it easy to change the configure package. > > At the moment it is extremely sub-optimal, with endless merge > > conflicts resulting solely from the need to have the ticket branch > > that changes configure.ac, spkg-configure.m4 files, and its friends, > > such as sage/big/sage-env* things coming with its configure "package" > > tarball. This leaves tickets of this sort languishing for months, as > > it's not really possible to get them in sync with the ongoing release > > process (already due to configure tarballs coming with artificial > > version numbers, that should not clash etc). > > > > I continue to think that the release process should not require > > updates of configure package. Sage seems to unique in its insistence > > on the release process depending on the autogenerated dumps... > > One very easily solution to this, and I know some people will shudder > (myself included) is to simply include the configure script in the > repository. Yes, it's generated. But there's precedent for this. well, it would make running patchbots easier, but I don't see how this would help the release management, as auto-merging would be still be out of the question. The auto-generated crud is simply unmergable, by and large. The patchbots are largely an orthogonal issue here. > Among other projects, CPython includes their configure script in the > repository. This allows developers to get right up-and-running with > ./configure without having to fuss with the nuances of different > versions of autotools on different systems. Only developers actually > working on configure issues need to worry about that. It's quite > convenient. > > > > > On Tue, Jun 11, 2019, 02:28 Kwankyu Lee wrote: > > >> > > >> > > >> > > >> On Monday, June 10, 2019 at 4:40:52 PM UTC+9, Sébastien Labbé wrote: > > >>> > > >>> Testing the following optional and external packages: > > >>> > > >>> External software detected for doctesting: > > >>> ffmpeg,graphviz,gurobi,imagemagick,internet,latex,pandoc > > >>> > > >>> Using > > >>> --optional=4ti2,bliss,cbc,ccache,cmake,cryptominisat,dot2tex,e_antic,external,glucose,latte_int,lidia,lrslib,memlimit,mpir,ninja_build,normaliz,notedown,openssl,pandoc_attributes,pycosat,pynormaliz,python2,qhull,rst2ipynb,sage,topcom > > >>> External software to be detected: > > >>> cplex,ffmpeg,graphviz,gurobi,imagemagick,internet,latex,macaulay2,magma,maple,mathematica,matlab,octave,pandoc,scilab > > >>> > > >>> I get: > > >>> > > >>> -- > > >>> sage -t --long src/sage/databases/findstat.py # 8 doctests failed > > >>> sage -t --long src/sage/combinat/designs/ext_rep.py # 1 doctest failed > > >>> -- > > >>> > > >>> and only the following is reproducible: > > >>> > > >>> -- > > >>> sage -t --long src/sage/databases/findstat.py # 8 doctests failed > > >>> -- > > >>> > > >>> Follow up at https://trac.sagemath.org/ticket/25536 > > >> > > >> > > >> I see Sage has been much improved in this respect. Thank you for your > > >> persistent efforts! > > >> > > >> -- > > >> You received this message because you are subscribed to the Google > > >> Groups "sage-release" group. > > >> To unsubscribe from this group and stop receiving emails from it, send > > >> an email to sage-release+unsubscr...@googlegroups.com. > > >> To post to this group, send email to sage-release@googlegroups.com. > > >> Visit this group at https://groups.google.com/group/sage-release. > > >> To view this discussion on the web visit > > >> https://groups.google.com/d/msgid/sage-release/35dc58d0-763e-4da4-b1c4-e8f981665b49%40googlegroups.com. > > >> For more options, visit https://groups.google.com/d/optout. > > > > > > -- > > > You received this message because you are subscribed to the Google Groups > > > "sage-release" group. > > > To unsubscribe from this group and stop receiving emails from it, send an > > > email to sage-release+unsubscr...@googlegroups.com. > > > To post to this group, send email to sage-release@googlegroups.com. > > > Visit this group at https://groups.google.com/group/sage-release. > > > To view this discussion on the web visit > > >
Re: [sage-release] Re: Sage 8.8.rc0 released
On Wed, Jun 12, 2019 at 11:45 AM Dima Pasechnik wrote: > > On Tue, Jun 11, 2019 at 7:52 AM E. Madison Bray wrote: > > > > I don't know why the rush to call this a release candidate when there are > > clearly still several outstanding problems to be resolved before the next > > release. > > > > As usual absolutely zero communication of a plan or coordination with the > > community. > > > my ongoing complaint is the lack of progress on the improving the > release procedures to make it easy to change the configure package. > At the moment it is extremely sub-optimal, with endless merge > conflicts resulting solely from the need to have the ticket branch > that changes configure.ac, spkg-configure.m4 files, and its friends, > such as sage/big/sage-env* things coming with its configure "package" > tarball. This leaves tickets of this sort languishing for months, as > it's not really possible to get them in sync with the ongoing release > process (already due to configure tarballs coming with artificial > version numbers, that should not clash etc). > > I continue to think that the release process should not require > updates of configure package. Sage seems to unique in its insistence > on the release process depending on the autogenerated dumps... One very easily solution to this, and I know some people will shudder (myself included) is to simply include the configure script in the repository. Yes, it's generated. But there's precedent for this. Among other projects, CPython includes their configure script in the repository. This allows developers to get right up-and-running with ./configure without having to fuss with the nuances of different versions of autotools on different systems. Only developers actually working on configure issues need to worry about that. It's quite convenient. > > On Tue, Jun 11, 2019, 02:28 Kwankyu Lee wrote: > >> > >> > >> > >> On Monday, June 10, 2019 at 4:40:52 PM UTC+9, Sébastien Labbé wrote: > >>> > >>> Testing the following optional and external packages: > >>> > >>> External software detected for doctesting: > >>> ffmpeg,graphviz,gurobi,imagemagick,internet,latex,pandoc > >>> > >>> Using > >>> --optional=4ti2,bliss,cbc,ccache,cmake,cryptominisat,dot2tex,e_antic,external,glucose,latte_int,lidia,lrslib,memlimit,mpir,ninja_build,normaliz,notedown,openssl,pandoc_attributes,pycosat,pynormaliz,python2,qhull,rst2ipynb,sage,topcom > >>> External software to be detected: > >>> cplex,ffmpeg,graphviz,gurobi,imagemagick,internet,latex,macaulay2,magma,maple,mathematica,matlab,octave,pandoc,scilab > >>> > >>> I get: > >>> > >>> -- > >>> sage -t --long src/sage/databases/findstat.py # 8 doctests failed > >>> sage -t --long src/sage/combinat/designs/ext_rep.py # 1 doctest failed > >>> -- > >>> > >>> and only the following is reproducible: > >>> > >>> -- > >>> sage -t --long src/sage/databases/findstat.py # 8 doctests failed > >>> -- > >>> > >>> Follow up at https://trac.sagemath.org/ticket/25536 > >> > >> > >> I see Sage has been much improved in this respect. Thank you for your > >> persistent efforts! > >> > >> -- > >> You received this message because you are subscribed to the Google Groups > >> "sage-release" group. > >> To unsubscribe from this group and stop receiving emails from it, send an > >> email to sage-release+unsubscr...@googlegroups.com. > >> To post to this group, send email to sage-release@googlegroups.com. > >> Visit this group at https://groups.google.com/group/sage-release. > >> To view this discussion on the web visit > >> https://groups.google.com/d/msgid/sage-release/35dc58d0-763e-4da4-b1c4-e8f981665b49%40googlegroups.com. > >> For more options, visit https://groups.google.com/d/optout. > > > > -- > > You received this message because you are subscribed to the Google Groups > > "sage-release" group. > > To unsubscribe from this group and stop receiving emails from it, send an > > email to sage-release+unsubscr...@googlegroups.com. > > To post to this group, send email to sage-release@googlegroups.com. > > Visit this group at https://groups.google.com/group/sage-release. > > To view this discussion on the web visit > > https://groups.google.com/d/msgid/sage-release/CAOTD34bCgqGe6UCE2eoDVq%2Bmg%2BW8f3nnRrBKztutSLvT%2BjicdQ%40mail.gmail.com. > > For more options, visit https://groups.google.com/d/optout. > > -- > You received this message because you are subscribed to the Google Groups > "sage-release" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sage-release+unsubscr...@googlegroups.com. > To post to this group, send email to sage-release@googlegroups.com. > Visit this group at https://groups.google.com/group/sage-release. >
Re: [sage-release] Re: Sage 8.8.rc0 released
On Tue, Jun 11, 2019 at 7:52 AM E. Madison Bray wrote: > > I don't know why the rush to call this a release candidate when there are > clearly still several outstanding problems to be resolved before the next > release. > > As usual absolutely zero communication of a plan or coordination with the > community. > my ongoing complaint is the lack of progress on the improving the release procedures to make it easy to change the configure package. At the moment it is extremely sub-optimal, with endless merge conflicts resulting solely from the need to have the ticket branch that changes configure.ac, spkg-configure.m4 files, and its friends, such as sage/big/sage-env* things coming with its configure "package" tarball. This leaves tickets of this sort languishing for months, as it's not really possible to get them in sync with the ongoing release process (already due to configure tarballs coming with artificial version numbers, that should not clash etc). I continue to think that the release process should not require updates of configure package. Sage seems to unique in its insistence on the release process depending on the autogenerated dumps... Dima > On Tue, Jun 11, 2019, 02:28 Kwankyu Lee wrote: >> >> >> >> On Monday, June 10, 2019 at 4:40:52 PM UTC+9, Sébastien Labbé wrote: >>> >>> Testing the following optional and external packages: >>> >>> External software detected for doctesting: >>> ffmpeg,graphviz,gurobi,imagemagick,internet,latex,pandoc >>> >>> Using >>> --optional=4ti2,bliss,cbc,ccache,cmake,cryptominisat,dot2tex,e_antic,external,glucose,latte_int,lidia,lrslib,memlimit,mpir,ninja_build,normaliz,notedown,openssl,pandoc_attributes,pycosat,pynormaliz,python2,qhull,rst2ipynb,sage,topcom >>> External software to be detected: >>> cplex,ffmpeg,graphviz,gurobi,imagemagick,internet,latex,macaulay2,magma,maple,mathematica,matlab,octave,pandoc,scilab >>> >>> I get: >>> >>> -- >>> sage -t --long src/sage/databases/findstat.py # 8 doctests failed >>> sage -t --long src/sage/combinat/designs/ext_rep.py # 1 doctest failed >>> -- >>> >>> and only the following is reproducible: >>> >>> -- >>> sage -t --long src/sage/databases/findstat.py # 8 doctests failed >>> -- >>> >>> Follow up at https://trac.sagemath.org/ticket/25536 >> >> >> I see Sage has been much improved in this respect. Thank you for your >> persistent efforts! >> >> -- >> You received this message because you are subscribed to the Google Groups >> "sage-release" group. >> To unsubscribe from this group and stop receiving emails from it, send an >> email to sage-release+unsubscr...@googlegroups.com. >> To post to this group, send email to sage-release@googlegroups.com. >> Visit this group at https://groups.google.com/group/sage-release. >> To view this discussion on the web visit >> https://groups.google.com/d/msgid/sage-release/35dc58d0-763e-4da4-b1c4-e8f981665b49%40googlegroups.com. >> For more options, visit https://groups.google.com/d/optout. > > -- > You received this message because you are subscribed to the Google Groups > "sage-release" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to sage-release+unsubscr...@googlegroups.com. > To post to this group, send email to sage-release@googlegroups.com. > Visit this group at https://groups.google.com/group/sage-release. > To view this discussion on the web visit > https://groups.google.com/d/msgid/sage-release/CAOTD34bCgqGe6UCE2eoDVq%2Bmg%2BW8f3nnRrBKztutSLvT%2BjicdQ%40mail.gmail.com. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "sage-release" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-release+unsubscr...@googlegroups.com. To post to this group, send email to sage-release@googlegroups.com. Visit this group at https://groups.google.com/group/sage-release. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-release/CAAWYfq1bWWce%2B-tN7V%2B-E7DGAFOGvu3L7san5WDR-M6p%2B0bA2A%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.