April 2016 - petsc-dev - mailman3.cels.anl.gov

Open PRs to be merged for release 3.7
by Lisandro Dalcin 24 Apr '16

24 Apr '16

Satish, I have two open PRs ready to be merged into master (already merged into next). As you are in charge of making the release, I did not want to step on you, so I'll not make the merge into master until you acknowledge. Or just merge them yourself through the Bitbucket interface. -- Lisandro Dalcin ============ Research Scientist Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) Extreme Computing Research Center (ECRC) King Abdullah University of Science and Technology (KAUST) http://ecrc.kaust.edu.sa/ 4700 King Abdullah University of Science and Technology al-Khawarizmi Bldg (Bldg 1), Office # 0109 Thuwal 23955-6900, Kingdom of Saudi Arabia http://www.kaust.edu.sa Office Phone: +966 12 808-0459

1 0

Re: [petsc-dev] plans for PETSc release
by Lisandro Dalcin 24 Apr '16

24 Apr '16

On 24 April 2016 at 16:06, Jed Brown <jed(a)jedbrown.org> wrote: > Okay, but I think we should never do a release this way again. We > should finish all merging to 'master', tag the release on 'master', then > fast-forward 'maint' and rewind 'next'. I'm with Jed. IMHO, that's the right workflow to make a new release. -- Lisandro Dalcin ============ Research Scientist Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) Extreme Computing Research Center (ECRC) King Abdullah University of Science and Technology (KAUST) http://ecrc.kaust.edu.sa/ 4700 King Abdullah University of Science and Technology al-Khawarizmi Bldg (Bldg 1), Office # 0109 Thuwal 23955-6900, Kingdom of Saudi Arabia http://www.kaust.edu.sa Office Phone: +966 12 808-0459

1 0

Re: [petsc-dev] plans for PETSc release
by Jed Brown 24 Apr '16

24 Apr '16

Satish Balay <balay(a)mcs.anl.gov> writes: > Master currently doesn't work as you describe. Why doesn't it work that way? That was the philosophy when we adopted this branching model years ago, it works reliably for many other projects, and I thought it worked for us when we used it that way. Did something change? > To me - currently we are at RC [yeah - witout a change in > petscversion.h or a tag] - and RC to RELEASE should be via maint > workflow - hence this update to maint. Why should RC to release "be via maint workflow"? I'd also argue we're clearly not at RC because new features (currently in 'next') are still being merged. After changing 'maint', you still advised Lisandro to merge new features to 'next' so that they could later graduate for the release. If any of those changes affect the API, then people on 'maint' have gotten an API that will never be released. This breaks our promises with respect to 'maint'. The whole point of that branch is to have something stable that users can update worry-free. > I didn't intend to be rude to our 'maint' users. Ideally there should > have been a tracking branch maint-3.6 always - so users who whish to > time the switch from 3.6 to 3.7 do this on their terms [and not > implicitly via maint]. > > I'll make an annoucement to maint users about switching to maint-3.6 > branch. Okay, but I think we should never do a release this way again. We should finish all merging to 'master', tag the release on 'master', then fast-forward 'maint' and rewind 'next'. As for pending merges, what is the status of these branches (in 'next', but not yet in 'master')? barry/fix-some-clang-warnings knepley/feature-fe-nonaffine knepley/fix-mem-init pr329/master/Fande-Kong/matpartitioning-hierarch pr359/master/Fande-Kong/fix-mat-convert-nest-aij tisaac/dmp4est-feature-mapped-coordinates tisaac/fix-ex3-coords

1 0

Re: [petsc-dev] plans for PETSc release
by Lisandro Dalcin 24 Apr '16

24 Apr '16

On 23 April 2016 at 22:54, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > >> On Apr 23, 2016, at 11:36 AM, Lisandro Dalcin <dalcinl(a)gmail.com> wrote: >> >> On 22 April 2016 at 23:24, Barry Smith <bsmith(a)mcs.anl.gov> wrote: >>> >>> So the current plan is that Satish will make tonights test run on master and then tomorrow (Saturday) he will make a new maint branch off of master. At that point you should only be adding fixes to the maint branch and not new features. >>> >> >> I have my TS work for 2nd order systems, as well as a BDF >> implementation that is almost ready (not yet in a branch, but I can >> put it in). Any chance we can include these new features? > > If you are willing to babysit them in maint (i.e. fix immediately any problems with them that come up in maint) then go ahead and get them into maint ASAP. Satish will likely make the tarball on Monday. > Yes, Barry, I'll commit to maintaining these new features. Moreover, there are a few things related to TSEvent support that should be done better, however these changes will not require API changes. Satish, I'll finish my work today and let you know ASAP about the two branches we should merge. -- Lisandro Dalcin ============ Research Scientist Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) Extreme Computing Research Center (ECRC) King Abdullah University of Science and Technology (KAUST) http://ecrc.kaust.edu.sa/ 4700 King Abdullah University of Science and Technology al-Khawarizmi Bldg (Bldg 1), Office # 0109 Thuwal 23955-6900, Kingdom of Saudi Arabia http://www.kaust.edu.sa Office Phone: +966 12 808-0459

1 0

message for petsc git users - using 'maint' branch
by Satish Balay 24 Apr '16

24 Apr '16

Dear PETSc 'maint' branch users, In anticipation of the next 3.7 release of petsc - I've updated 'maint' branch in petsc git repository to 3.7 pre-release changes. If you 'git pull' on maint branch - you would get these 3.7 pre-release changes. To stay on 3.6 - you can switch to 'maint-3.6' branch. git fetch git checkout maint-3.6 I apologize for any inconvenience this has caused. Satish

1 0

Re: [petsc-dev] plans for PETSc release
by Satish Balay 23 Apr '16

23 Apr '16

On Sat, 23 Apr 2016, Jed Brown wrote: > Satish Balay <balay(a)mcs.anl.gov> writes: > > > Ok - the master tests ran overnight > > > > I've switched the day builds to maint. [nightbuilds are reverted to next] > > > > Also created maint-3.6 branch off maint - and updated maint with master. > > Uh, then people that have been tracking 'maint' get something backward > incompatible before a release has been tagged or announced. I think > that's rude. How did this become the release process? I thought we > declared a merge deadline for 'master', feature freeze period, then > tagged the release on 'master'. That way people tracking 'maint' only > get the new release changes after the release has been made. I'm sure you'll find faults with my answer - but here it goes.. Master currently doesn't work as you describe. To me - currently we are at RC [yeah - witout a change in petscversion.h or a tag] - and RC to RELEASE should be via maint workflow - hence this update to maint. I didn't intend to be rude to our 'maint' users. Ideally there should have been a tracking branch maint-3.6 always - so users who whish to time the switch from 3.6 to 3.7 do this on their terms [and not implicitly via maint]. I'll make an annoucement to maint users about switching to maint-3.6 branch. Satish

1 0

Re: [petsc-dev] plans for PETSc release
by Jed Brown 23 Apr '16

23 Apr '16

Satish Balay <balay(a)mcs.anl.gov> writes: > Ok - the master tests ran overnight > > I've switched the day builds to maint. [nightbuilds are reverted to next] > > Also created maint-3.6 branch off maint - and updated maint with master. Uh, then people that have been tracking 'maint' get something backward incompatible before a release has been tagged or announced. I think that's rude. How did this become the release process? I thought we declared a merge deadline for 'master', feature freeze period, then tagged the release on 'master'. That way people tracking 'maint' only get the new release changes after the release has been made.

1 0

Re: [petsc-dev] plans for PETSc release
by Balay, Satish 23 Apr '16

23 Apr '16

Wrt next - some of the tests start at midnight, some at 2am etc. Then at 9am the logs are collected and made available on dashboard. Similarly day builds (master now maint) start at noon ...... Satish ________________________________ From: Barry Smith<mailto:[email protected]> Sent: ‎4/‎23/‎2016 3:44 PM To: Balay, Satish<mailto:[email protected]> Cc: Lisandro Dalcin<mailto:[email protected]>; petsc-dev<mailto:[email protected]> Subject: Re: [petsc-dev] plans for PETSc release > On Apr 23, 2016, at 3:33 PM, Balay, Satish <balay(a)mcs.anl.gov> wrote: > > I think the first thing to do is get into next by midnight cst and look at merging to maint tomorrow. Agreed a round of next tonight before maint is best. But why midnight, shouldn't it be 9pm central time? > > Satish > From: Barry Smith > Sent: ‎4/‎23/‎2016 2:54 PM > To: Lisandro Dalcin > Cc: petsc-dev > Subject: Re: [petsc-dev] plans for PETSc release > > > > On Apr 23, 2016, at 11:36 AM, Lisandro Dalcin <dalcinl(a)gmail.com> wrote: > > > > On 22 April 2016 at 23:24, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > >> > >> So the current plan is that Satish will make tonights test run on master and then tomorrow (Saturday) he will make a new maint branch off of master. At that point you should only be adding fixes to the maint branch and not new features. > >> > > > > I have my TS work for 2nd order systems, as well as a BDF > > implementation that is almost ready (not yet in a branch, but I can > > put it in). Any chance we can include these new features? > > If you are willing to babysit them in maint (i.e. fix immediately any problems with them that come up in maint) then go ahead and get them into maint ASAP. Satish will likely make the tarball on Monday. > > Barry > > > > > > > -- > > Lisandro Dalcin > > ============ > > Research Scientist > > Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) > > Extreme Computing Research Center (ECRC) > > King Abdullah University of Science and Technology (KAUST) > > http://ecrc.kaust.edu.sa/ > > > > 4700 King Abdullah University of Science and Technology > > al-Khawarizmi Bldg (Bldg 1), Office # 0109 > > Thuwal 23955-6900, Kingdom of Saudi Arabia > > http://www.kaust.edu.sa > > > > Office Phone: +966 12 808-0459

1 0

Re: [petsc-dev] plans for PETSc release
by Barry Smith 23 Apr '16

23 Apr '16

> On Apr 23, 2016, at 3:33 PM, Balay, Satish <balay(a)mcs.anl.gov> wrote: > > I think the first thing to do is get into next by midnight cst and look at merging to maint tomorrow. Agreed a round of next tonight before maint is best. But why midnight, shouldn't it be 9pm central time? > > Satish > From: Barry Smith > Sent: ‎4/‎23/‎2016 2:54 PM > To: Lisandro Dalcin > Cc: petsc-dev > Subject: Re: [petsc-dev] plans for PETSc release > > > > On Apr 23, 2016, at 11:36 AM, Lisandro Dalcin <dalcinl(a)gmail.com> wrote: > > > > On 22 April 2016 at 23:24, Barry Smith <bsmith(a)mcs.anl.gov> wrote: > >> > >> So the current plan is that Satish will make tonights test run on master and then tomorrow (Saturday) he will make a new maint branch off of master. At that point you should only be adding fixes to the maint branch and not new features. > >> > > > > I have my TS work for 2nd order systems, as well as a BDF > > implementation that is almost ready (not yet in a branch, but I can > > put it in). Any chance we can include these new features? > > If you are willing to babysit them in maint (i.e. fix immediately any problems with them that come up in maint) then go ahead and get them into maint ASAP. Satish will likely make the tarball on Monday. > > Barry > > > > > > > -- > > Lisandro Dalcin > > ============ > > Research Scientist > > Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) > > Extreme Computing Research Center (ECRC) > > King Abdullah University of Science and Technology (KAUST) > > http://ecrc.kaust.edu.sa/ > > > > 4700 King Abdullah University of Science and Technology > > al-Khawarizmi Bldg (Bldg 1), Office # 0109 > > Thuwal 23955-6900, Kingdom of Saudi Arabia > > http://www.kaust.edu.sa > > > > Office Phone: +966 12 808-0459

1 0

Re: [petsc-dev] plans for PETSc release
by Balay, Satish 23 Apr '16

23 Apr '16

I think the first thing to do is get into next by midnight cst and look at merging to maint tomorrow. Satish ________________________________ From: Barry Smith<mailto:[email protected]> Sent: ‎4/‎23/‎2016 2:54 PM To: Lisandro Dalcin<mailto:[email protected]> Cc: petsc-dev<mailto:[email protected]> Subject: Re: [petsc-dev] plans for PETSc release > On Apr 23, 2016, at 11:36 AM, Lisandro Dalcin <dalcinl(a)gmail.com> wrote: > > On 22 April 2016 at 23:24, Barry Smith <bsmith(a)mcs.anl.gov> wrote: >> >> So the current plan is that Satish will make tonights test run on master and then tomorrow (Saturday) he will make a new maint branch off of master. At that point you should only be adding fixes to the maint branch and not new features. >> > > I have my TS work for 2nd order systems, as well as a BDF > implementation that is almost ready (not yet in a branch, but I can > put it in). Any chance we can include these new features? If you are willing to babysit them in maint (i.e. fix immediately any problems with them that come up in maint) then go ahead and get them into maint ASAP. Satish will likely make the tarball on Monday. Barry > > > -- > Lisandro Dalcin > ============ > Research Scientist > Computer, Electrical and Mathematical Sciences & Engineering (CEMSE) > Extreme Computing Research Center (ECRC) > King Abdullah University of Science and Technology (KAUST) > http://ecrc.kaust.edu.sa/ > > 4700 King Abdullah University of Science and Technology > al-Khawarizmi Bldg (Bldg 1), Office # 0109 > Thuwal 23955-6900, Kingdom of Saudi Arabia > http://www.kaust.edu.sa > > Office Phone: +966 12 808-0459

1 0