Dealii: deal.II Release 9.0 5/11/2018

Created on 4 Apr 2018 · 153Comments · Source: dealii/dealii

Documentation version v9.0.0

Open tasks:

Update dealii in Ubuntu/Debian (Matthias)
Update dealii virtualbox images (Timo)
Send a note to the Facebook and Google+ accounts (Wolfgang/Timo)

Merge history:

final: e9f6e3ab9d495d9c8a14b11e2d4a374b6acc6fa1
- #6580 (take over #6567)
- #6579 (take over #6577)
- #6575 (take over #6573, #6574)
- #6571 (take over #6569)
- #6570 (take over #6564)
- #6563 (take over #6542)
- #6561 (take over #6555)
- #6560 (take over #6553)
rc5: 6ac831ab500cc6eca5cf66b9eaad47419d2a226a
- #6551 (take over #6540)
rc4: 27f62c0b80
- #6549 (take over #6544, #6543, #6541)
- #6548 (take over #6537, #6538
- #6547 (take over #6535, #6536)
- #6534 (take over #6533)
- #6521 (take over #6297)
- #6516 (take over #6291, #6511, #6513)
- #6515 (take over #6508, #6519)
- #6514 (take over #6393)
rc3: 3975bada23997cdc376a247bd55295a4eedbfca8
- #6509 (take over #6475)
- #6507 (take over #6499)
- #6506 (take over #6498)
- #6505 (take over #6496)
- #6504 (take over #6495)
- #6503 (take over #6491)
- #6502 (take over #6489)
- #6501 (take over #6483)
- #6500 (take over #6480)
- #6492
- #6485
rc2: bfe525dc113b40eeea79a197aa5eebaac9566f1f
- #6482
- #6479 (take over #6474)
- #6478 (take over #6476)
- #6477
rc1: 98208d44b01e6b60cc31ee08514633300c11fcfa

Testsuite results for v9.0.0-rc5-6-ge549f8f0bf e549f8f0bf

| Host | Configuration | Commit | Build errors | Build warnings | Failing tests | Passing tests |
| - | - | - | :-: | :-: | :-: | :-: |
| tester | GNU-7.3.0-dealii-9.0-unity_build | e549f8f0bf | 0 | 0 | 0 | 0 |
| tester | GNU-7.3.0-dealii-9.0-all_components | e549f8f0bf | 0 | 0 | 0 | 0 |
| tester | GNU-7.3.0-dealii-9.0-minimal_bundled | e549f8f0bf | 0 | 0 | 0 | 7090 |
| tester | GNU-4.9.4-dealii-9.0-autodetection | e549f8f0bf | 0 | 0 | 0 | 10093 |
| tester | GNU-6.4.0-dealii-9.0-autodetection | e549f8f0bf | 0 | 0 | 0 | 10157 |
| tester | GNU-7.3.0-dealii-9.0-autodetection | e549f8f0bf | 0 | 0 | 0 | 10157 |
| tester | GNU-8.1.0-dealii-9.0-autodetection | e549f8f0bf | 0 | 0 | 0 | 10157 |
| tester | Clang-4.0.1-dealii-9.0-autodetection | e549f8f0bf | 0 | 0 | 0 | 10139 |
| tester | Clang-5.0.1-dealii-9.0-autodetection | e549f8f0bf | 0 | 0 | 0 | 10139 |
| tester | Clang-6.0.0-dealii-9.0-autodetection | e549f8f0bf | 0 | 0 | 0 | 10139 |
| tester | GNU-7.3.0-dealii-9.0-petsc_complex | e549f8f0bf | 0 | 0 | 1 | 9850 |
| tester | GNU-7.3.0-dealii-9.0-64bit_indices | e549f8f0bf | 0 | 0 | 4 | 10119 |
| tester | GNU-7.3.0-dealii-9.0-optimized | e549f8f0bf | 0 | 0 | 20 | 10137 |
| tester | Clang-6.0.0-dealii-9.0-optimized | e549f8f0bf | 0 | 0 | 11 | 10128 |
| simserv04 | GNU-5.3.0-dealii-9.0-autodetection | efe2dd0 | 0 | 0 | 0 | 9051 |
| simserv04 | Clang-3.7.0-dealii-9.0-autodetection | efe2dd0 | 0 | 0 | 2 | 9049 |
| simserv02 | Intel-18.0.1.20171018 | e549f8f | 0 | 0 | 4 | 8405 |
| simserv02 | Intel-19.0.0.20180317 | e549f8f | 0 | 19 | 115 | 8316 |
| davyddenubuntu | GNU-7.3.0-dealii-9.0-^openmpi^openblas | efe2dd00 | 0 | 0 | 0 | 10147 |
| CUDA8 | GNU-5.4.0-dealii-9.0 | efe2dd0 | 0 | 80 | 2 | 7068 |
| CUDA9 | GNU-5.4.0-dealii-9.0 | efe2dd0 | 0 | 0 | 2 | 7068 |
| node27 | GNU-6.1.0-dealii-9.0 | efe2dd0 | 0 | 0 | 3 | 9183 |
| node27 | GNU-6.1.0-dealii-9.0-nothread | efe2dd0 | 0 | 0 | 3 | 9163 |

Checklist for release-tasks

Release preparation steps:

[x] -3 Send out notice to the mailinglist if anyone has outstanding issues or patches.
[x] -2 Check all issues and pull-request on github. Create a list of things to fix before the release.
[x] -1 Update the list of publications. (Send standard e-mail)
[x] -1a Start release paper process with email to all current (principal) developers
[x] -1b Update the authors list on the homepage authors.html
[x] -1c Update the list of contributors in the release paper
[x] 0 Do some basic testing as always
[x] 0a Rerun cppcheck across the source files
[x] 0b Check for doxygen formatting problems
[x] 0c Fix formatting bugs reported by various command line queries
[x] 0d Check that doxygen produces no errors when generating offline docs.
[x] 0e Run the copyright script
[x] 0f Update Trilinos and PETSc to the latest release on the tester and update documentation of supported version.
[x] 0g Run codespell

Release steps:

[x] * Nominate a "release officer" that is in charge of (almost all) of the following steps
[x] 1 Create new changes file
[x] 1a Adjust doc/users/doxygen.html to link to the to-be-created TAG file.
[x] 4 Create a release branch
[x] 5 Update the VERSION from 9.0.0-pre to 9.1.0-pre on master
[x] 6 Update the official tester to test on the release branch dealii-9.0
[x] 6a On mainline, update deal.II dependencies of examples
[x] 7 Create an AUTHORS file
[x] 7a Send prerelease announcement to [email protected]
[x] 8 Remove unfinished tutorial programs.
[x] 9 Write up a short release note and save it as announce-9.0
[x] 9a Write the release paper
[x] 10 Test the branch thoroughly
[x] 10a Do the same for the code gallery examples
[x] 11 Create documentation and verify
[x] 12 Push the button
- Change VERSION to 9.0.0 for a release and to 9.0.0-rc1 for a pre-release
- Create a signed tag of exactly this commit
- Push changes to github
[x] 13 Create the tar files
- Download the source tarball from github, verify that its contents is what we expect
- Cryptographically sign it (detached, armored signature)
- Create full offline_doc tarball and sign it
[x] 14a Create a pre-release on github and upload files (and detached signatures)
[x] 15 Generate the documentation on the webserver.
[x] 16 Adjust header.include on homepage
[x] 17a Ask Luca for Mac packages
[x] 17b Update dealii in Spack
[x] 17c Update dealii in Candi
[x] 17d Update dealii in Gentoo Science overlay
[ ] 17e Update dealii in Ubuntu/Debian
[x] 17f Update dealii in Archlinux AUR
[ ] 17g Update dealii virtualbox images
[x] 18 Finish writing the announcement
[x] 18b Adjust news.html in the deal.II homepage repo
[x] 19 Announce the release via e-mail
[ ] 19a Send a note to the Facebook and Google+ accounts
[x] 19b Update the deal.II wikipedia page
[x] 19d Send thank-you email and list the people already referenced in step -1b above
[x] 20 Celebrate!

Release

Source

tamiko

🎉6 😄1

Most helpful comment

On 05/12/2018 07:32 AM, David Wells wrote:

I have a revised preprint that gets rid of the marginalia and some of the
lines that extend into the margins,

I've gone over the preprint again and edited it here and there as appropriate.
Should we submit it to JNM again, as we did for the previous ones?

bangerth on 14 May 2018

👍4

All 153 comments

Do we have a deadline for when we will create a release branch and only accept bugfixes?

drwells on 9 Apr 2018

End of April?

davydden on 9 Apr 2018

I guess @drwells also commented on being more conservative on master before creating the branch. In my opinion it would make sense to have that deadline a bit before April 27. I would have said that all non-bugfix PRs should be posted by April 20 to have some time to review and discuss. But even that seems ambitious if we target a "release" by the end of the month.

Furthermore, we have a few substantial issues left on our list. We should decide this week which of those we want to finish and which ones to postpone because it seems unrealistic to address all 30 open issues for the 9.0 release.

kronbichler on 9 Apr 2018

Can we switch to a mode where we now only accept bug fixes, doc changes, and, possibly, new tutorial programs? We really need to get to a point where we can cut the release branch.

What functionality is currently still pending? What is lacking regarding the manifold changes that still needs to go in?

bangerth on 23 Apr 2018

Also, if everyone could go through the milestone list at https://github.com/dealii/dealii/milestone/5 and think about whether their favorite issue really needs to be in 9.0 or could slip to 9.1, that would be great!

bangerth on 23 Apr 2018

Can we switch to a mode where we now only accept bug fixes, doc changes, and, possibly, new tutorial programs? We really need to get to a point where we can cut the release branch.

I have #5667 which is 75% through. I will post the second to last big patch (everything apart from FEFaceEvaluation) tonight. I apologize for being late, but it was a lot to review and we did our best to break things down which delayed things since we started in February.

kronbichler on 23 Apr 2018

I would like to merge one more PR for cuda. It will be ready tonight, the code is ready but the computer were the code is, was down during the week-end so I couldn't create the PR.

Rombur on 23 Apr 2018

@bangerth

Can we switch to a mode where we now only accept bug fixes, doc changes, and, possibly, new tutorial programs?

I think we don't have to switch to that mode, we simply need to start a 9.0-pre branch and cherry-pick simple PRs there while keep reviewing and merging things into the master as usual.

davydden on 23 Apr 2018

What about we fork off the release branch this Friday and create a first release candidate? This should give everyone enough time for their last-minute inclusions :-)

After that I suggest to tag the second release candidate Friday in a week. Depending on how things calmed down in that week we can make the final releas shortly after :-)

tamiko on 23 Apr 2018

👍4

I like the idea of forking at the end of the week. Right now, there are still too many open PRs that would need to be cherry-picked by hand, so there is little benefit to forking now. But let's see where we are on Friday.

bangerth on 24 Apr 2018

👍1

I have compiled a handy check list for our release tasks (see updated first post).
As always, we need exactly 20 steps for a release ;-)

tamiko on 24 Apr 2018

As always, we need exactly 20 steps for a release ;-)

:-)

tjhei on 24 Apr 2018

As always, we need exactly 20 steps for a release ;-)

Actually, 42 is a pretty nice number.

masterleinad on 24 Apr 2018

😄1

Remarkably, if you ignore the negative numbers (and count the bullets) we do actually have 42 steps.

drwells on 24 Apr 2018

@tamiko -- thanks for converting the release tasklist file into this github issue. This is more easy to work with. If you wanted to convert the file in a way so that next time we only need to copy-paste it into a github issue, then that would be ok as well!

bangerth on 25 Apr 2018

I realized the 42 thing shortly after writing that comment :grinning:
@bangerth Already part of the release-papers repo :smiley: https://github.com/dealii/release-papers/blob/master/release-tasks-github-list

tamiko on 25 Apr 2018

Depending on how much work we get done over the course of this day I plan to cut a release branch at around midnight UTC (that is 19:00 Central time). (Well, also depending on how much shouting there is to postpone ^^)

tamiko on 27 Apr 2018

👍1

There is still step-60 (missing basically only the results section) and a gmsh function I intended to add.

luca-heltai on 27 Apr 2018

If you push for a release today, that's not going to happen on my side, unless @GivAlz manages to get things into a final shape in step-60 by midnight your time.

luca-heltai on 27 Apr 2018

Well, then it is in a week, I guess.

tamiko on 27 Apr 2018

@luca-heltai Actually, I wanted to tag a pre-release, not a release. But let's wait with that for another week.

tamiko on 27 Apr 2018

If I'm the only one holding back, by all means let's call it a day, and go for a pre-release today.

I thought step-60 would have been a good thing to have, since it basically uses only new functionality, not treated in any other tutorial (MappingFEField, NonMatching, ParameterAcceptor, and others...)

luca-heltai on 27 Apr 2018

@luca-heltai We can also tag a release today (rc1) and merge the new tutorial step into rc2 next week Friday. That way we would have a feature-frozen release candidate for testing and you would have a week to finish the example step. I can also help Tuesday onwards if you wish :-)

tamiko on 27 Apr 2018

👍2

Let's wait before we remove gmsh though...

luca-heltai on 27 Apr 2018

@kronbichler also has a new tutorial https://github.com/dealii/dealii/pull/6291 for DG, but i see it's being tagged 9.1. I thought it was only a matter of merging that final matrix-free DG PR and rebasing the tutorial.

davydden on 27 Apr 2018

@kkormann and I can get that tutorial done until next week, but we need half a day (which is first next Wednesday for me) because we have not written the documentation yet, just the plain program. Let us see what others think about this.

kronbichler on 27 Apr 2018

@kkormann and I can get that tutorial done until next week

IMO one week does not break the deal (pun intended) 😄 so let's more-or-less freeze the features but not branch off yet as we have a couple of opened issues.

davydden on 27 Apr 2018

One week would be great: it would allow to finish the result section of step-60 and improve the tutorial overall! I noticed the NonMatching functions need a better documentation, which I would also write as it is functional for the tutorial! Thanks :)

GivAlz on 27 Apr 2018

This would also give me time to attend to #6172, so I vote with the crowd on this.

jppelteret on 27 Apr 2018

Guess next week it is then :-)

tamiko on 27 Apr 2018

It takes a substantial amount of time to go through the release task list. There is value to making the branch earlier rather than later, so that people can go through the list. We can always pull commits over to the branch.

Could I advocate for the following:

We work on merging all of the pending bug fix and doc fix PRs.
When they are in, we cut the branch (whenever that is).
We go through the release task list and do all of the work on the release branch as necessary.
If one or both of the tutorial programs happens to be ready before the release date, we can always pull that over without too much risk.

So the work for the moment is that we need to go through the open pull requests and issues that we know we need to get through.

bangerth on 27 Apr 2018

@bangerth

We work on merging all of the pending bug fix and doc fix PRs.

When they are in, we cut the branch (whenever that is).

I strongly agree. Further, I suggest, we do not merge anything else to master until we forked off.

tamiko on 28 Apr 2018

I also agree.

I have to fix a few failing tests due to #6300 (because SSE generates different output than AVX that I tested with) that I will commit later today: http://cdash.kyomu.43-1.org/viewTest.php?onlyfailed&buildid=1176

kronbichler on 28 Apr 2018

All, as far as I can see we are basically down to

a number of minor and major documentation updates that are still partially pending
(#6274, ~~#6184~~, #6183, #6172)
one obscure bug (#6229)
~~Two pending pull request (#6434, #6428)~~
Two new example steps (#6291, #6297)

If everyone is OK with it, I would like to:

suggest to embargo the main branch for the next 24h (only pull requests for above stuff gets merged).
tag a prerelease tomorrow at above time immediately after the two pending pull requests are merged (that is 5/4/2018 on or before 23:59 UTC ^^).
suggest we merge further documentation fixes on the release branch as they come in (over the next week).
suggest we wait for the actual release for step-59 and step-60 to be polished and merged to the release.

I would really like to freeze the library right now with most of the testsuite being exceptionally happy...
I don't see any problems for documentation and example steps being merged quite late in the release cycle.

tamiko on 4 May 2018

👍3

I have nominated e4483a2 for forking off. I will run bob with one configuration quickly to make sure that everything is in order.

tamiko on 4 May 2018

I posted a warning in the gitter chat that I should probably repeat here.

A possible problem with 9b5030eba6bf6b536037dab8e7d7a7f124fa3866 is that it changes (slightly) the 96 cell hyper shell in 3D. We have about 50 tests that rely on this grid, so I may have broken more tests due to this.

I'm looking into it now (I am rerunning the full test suite on my workstation).

drwells on 4 May 2018

@drwells I will wait with the ~~branching~~tagging until after the testsuite results.

tamiko on 4 May 2018

@dealii/dealii We have a release branch now \o/ I will tag a RC1 sometime tomorrow after the latest fallout (https://github.com/dealii/dealii/issues/6458) is fixed.

tamiko on 5 May 2018

the 96 cell hyper shell in 3D

this will have additional fall-out for ASPECT and its testsuite @bangerth @gassmoeller

tjhei on 5 May 2018

Out of curiousity I recompiled aspect and ran the test suite:

25% tests passed, 422 tests failed out of 562

with failures like

561: ******* Check .../output-zero_matrix/statistics                                                          
561: ******* All of 9 lines of diffs are:                                                                                    
561: ----------------                                                                                                        
561: ##18      <== 0 0.000000000000e+00 0.000000000000e+00 1000 9553 4221 8442 4 0 0 0 54 58 58 output-zero_matrix/solution/ 
solution-00000 1.01711092e-02 1.03408785e-02                                                                                 
561: ##18      ==> 0 0.000000000000e+00 0.000000000000e+00 1000 9553 4221 8442 4 0 0 0 50 58 58 output-zero_matrix/solution/solution-00000 1.01711092e-02 1.03408785e-02                                                                                 
561:                                                                                                                         
561: ##18      #:12  <== 54                                                                                                  
561: ##18      #:12  ==> 50                                                                                                  
561: @ Absolute error = 4.0000000000e+0, Relative error = 8.0000000000e-2

Should we revert this change? This is causing a massive amount of pain already.

drwells on 5 May 2018

Depends, how confident are you that the new version is better? We do not check for correctness of test results with the recent deal.II dev (our reference tester is running 8.5) so there is no inherent reason not to change something for deal.II 9.0, just because it changes things for aspect. Of course it would be bad if we later on discover the new version is worse than the existing one, but at least in the result that you show above the lower iteration count suggests the new geometry is actually better (and I get the same diff on my system). See also my comment about test results in aspect with deal-9.0 here: geodynamics/aspect#2185

Edit: I should have checked the diff you referred to beforehand. Do I understand correctly that you just replaced the manual vertex locations by the SphericalManifold, but in theory they should be the nearly identical (except for some roundoff errors)? Then I actually prefer the new version.

gassmoeller on 5 May 2018

@drwells I wouldn't be too worried about the hypershell update. We also have a week to assess how much fallout this created.

tamiko on 5 May 2018

@masterleinad @Rombur @davydden Please reconfigure all testers that report to Continous to test tag v9.0.0-r1, commit 98208d4 so that we can update above table :-)

tamiko on 5 May 2018

Will do on Monday

davydden on 5 May 2018

@gassmoeller I did not examine the quality of the shapes: they are different by more than roundoff errors but yes, the new points are computed by SphericalManifold.

drwells on 5 May 2018

I think the new grid is better: I posted a longer discussion in the aspect tracker.

drwells on 5 May 2018

@davydden Wise choice - v9.0.0-rc2 it should be by then ;-)

tamiko on 5 May 2018

@masterleinad @Rombur @davydden Please run all testing sites reporting to Continuous once on branch dealii-9.0, specifically commit 3975bad

tamiko on 7 May 2018

@Rombur What is the current state of CUDA8 for the release branch dealii-9.0?

tamiko on 7 May 2018

The current status is that we have problems with the testing machine :)

Rombur on 7 May 2018

Please run all testing sites reporting to Continuous once on branch dealii-9.0, specifically commit 3975bad

The last three characters make this an extremely funny hash name for a release tag... :laughing:

kronbichler on 7 May 2018

😄2

@kronbichler Jup!

tamiko on 7 May 2018

Please run all testing sites reporting to Continuous once on branch dealii-9.0

my openmpi+openblas tester only complained about gmsh/create_tria_02, which is failing everywhere... I guess @bangerth was right that we should have postponed the new functionality ;-( Let's see if https://github.com/dealii/dealii/pull/6508 fixed it...

davydden on 7 May 2018

my openmpi+openblas tester only complained about
gmsh/create_tria_02, which is failing everywhere... I guess
@bangerth was right that we should have postponed the new
functionality ;-(

The test and the functionality is fixed on master. I have high hopes for
rc4 to be down to 0 failing tests and no known regressions. :-)

tamiko on 7 May 2018

Do we have an ETA for rc4?

drwells on 7 May 2018

@drwells I would prefer to tag a rc4 on Wednesday and tag the release on Friday.

We have two fixes for rc4 that can go in after bob has tested current master once. I suggest to wait to tag rc4 until after step-59 and step-60 made it into master and the release branch. Hopefully we get that done by Wednesday :-)

What are your opinions?

tamiko on 7 May 2018

@tamiko #6291 and #6511 are already merged to master. It's now only a tiny update #6513 that is missing for step-59.

kronbichler on 7 May 2018

I don't know how close step-60 is so I do not have a strong opinion on deadlines (other than we should have both new steps included).

drwells on 7 May 2018

I don't know how close step-60 is

As far as I can tell, all the comment are addressed, @jppelteret wanted to give it another look as he previously requested changes. I think tagging rc4 on Wednesday and releasing on Friday is doable.

davydden on 7 May 2018

Yip, I've reviewed #6297 and I know that I've still got some documentation outstanding. The (now more concrete) plan for me is to finish #6172 tomorrow. Hopefully @drwells and I can come to some sensible resolution on #6473 as well.

jppelteret on 7 May 2018

If #6172 and #6437 don't happen today, I vote for postponing to 9.1. Both fall in the "nice to have, but not critical" category.

Re @davydden 's comment about the mkstemp issue -- we'll have to have some sort of debrief about our release process once we're finally done.

bangerth on 8 May 2018

👍1

@bangerth I thought we already postponed #6437 to 9.1, or should I have worked on that for 9.0?!

tamiko on 8 May 2018

I thought we already postponed #6437 to 9.1, or should I have worked on that for 9.0?!

I think he meant #6473 (last two numbers swapped)

davydden on 8 May 2018

@davydden That makes much more sense. :-)

tamiko on 8 May 2018

Bizarre that both numbers made sense. And yes, I swapped the numbers.

bangerth on 8 May 2018

@bangerth Above is my suggestion for a final set of merges for rc4. I think we shouldn't do much more.

tamiko on 8 May 2018

we also have those two https://github.com/dealii/dealii/issues/6274 and https://github.com/dealii/dealii/issues/6183 marked with 9.0 milestone.

davydden on 8 May 2018

Let's skip #6517. It's a nuisance that it doesn't work, but I don't think it's critical and we need to come to a conclusion that doesn't require us running the test suite again and again. It can wait for 9.1.

So the 4 pull requests you mention sound good to me.

bangerth on 8 May 2018

@bangerth I am happy to tag rc4 (around 1pm central time tomorrow) once all of these pull-requests are merged. This should give us solid 3 days to do some final testing before we tag the release.
(3 days allows potentially 6 full runs of the testsuite on tester/bob.)

tamiko on 8 May 2018

Let's postpone #6183 (glossary entry for distorted cells) to 9.1 -- same story: nice to have, not critical. (It's going to be outdated -- not ideal, but we can live with that).

I'd say let's also postpone whatever the remainder of #6274 (documentation of GridGenerator) is. It looks like almost everything was updated. Whatever else is left are probably marginal cases where either the documentation is not quite up to date or where we'd have to change code in non-trivial ways (with the consequent need to run tests). But I'd like to let @drwells make that call.

bangerth on 8 May 2018

👍2

I would like to attach TorusManifold objects by default. This is not our most popular GridGenerator function but I believe it is the last one whose behavior needs to be updated.

drwells on 8 May 2018

@dealii/dealii All, I will restart tester on the release branch now. If all is going well (i.e. the first 4 configuration pass), I pan to tag rc4 at around 1pm central time and restart again.
Have I missed any urgent last minute changes that should go into rc4?

tamiko on 8 May 2018

@dealii/dealii mhm I am actually tempted to wait a bit for #6537, #6538, #6536, #6535 and add them to the rc4 as well. (And after that bite at everyone who wants to have further PRs that are not critical bugfixes ^^). Opinions?

tamiko on 8 May 2018

👍1

I will, of course, lobby for #6537 and #6538. The other two adjust tests and should also be okay.

drwells on 8 May 2018

@luca-heltai What version of gmsh do you use and what features are configured? I cannot get the gmsh/create_tria_02 test to pass on any of my machines.

tamiko on 8 May 2018

634-heltai@turbolo: step-60[hot-fix-step-60] $ gmsh -info
Version : 3.0.6
License : GNU General Public License
Build OS : MacOSX
Build date : 20180505
Build host : turbolo
Build options : 64Bit Ann Bamg Bfgs Blas(Custom) Blossom C++11 Chaco DIntegration Dlopen GMP Gmm Kbipack Lapack(Custom) MPI MathEx Mesh Mmg3d NativeFileChooser Netgen ONELAB ONELABMetamodel OptHom Parser Plugins Post Solver TetGen/BR Tetgen1.5 Voro3D
Packaged by : heltai
Web site : http://gmsh.info
Mailing list : [email protected]

luca-heltai on 8 May 2018

@luca-heltai Ah, blas, netgen and tetgen support was missing.

tamiko on 8 May 2018

@luca-heltai It might be a good idea to eventually improve the FindGMSH.cmake script and query for supported modes (in case DEAL_II_ALLOW_PLATFORMINTROSPECTION is set to true) and verbosely tell if gmsh misses some crucual configuration.

tamiko on 8 May 2018

@tamiko, the thing is that I have no idea what modes we need... it is not really well documented what options work with what configuration. We tried a few, and went for using the most robust one. It does not say in the documentation that you need a certain set of features... I'll try to figure out something better for 9.1

luca-heltai on 8 May 2018

👍1

@tamiko @luca-heltai create_tria_02 still fails on my tester. blas is there in gmsh, but netgen and tetgen are not, i will try adding them to gmsh and see if it fixes the issue.

Meanwhile, @luca-heltai could you fix this https://github.com/dealii/dealii/issues/5643 ? I don't think gmsh dependency is documented, you could also mention what is known to be required for new functionality to work.

davydden on 8 May 2018

Meanwhile, @luca-heltai could you fix this #5643 ?

hm, it looks like it is documented now https://www.dealii.org/developer/index.html maybe just update this text with requirements on gmsh and close the issue?

davydden on 8 May 2018

i added a line about Gmsh https://github.com/dealii/dealii/pull/6541

@luca-heltai

Meanwhile, @luca-heltai could you fix this #5643 ?

5643 is actually about installation instructions, so it's not just what Gmsh needs to be build with. Nevermind my comment.

davydden on 8 May 2018

@dealii/dealii mhm I am actually tempted to wait a bit for #6537, #6538, #6536, #6535 and add them to the rc4 as well. (And after that bite at everyone who wants to have further PRs that are not critical bugfixes ^^). Opinions?

All four of these are now merged. Only the first changes any code, the rest are probably safe to pull over to the branch, though I don't think the last two really are critical. Either way, let's bring these over and then finally be done! We've all spent enough time on this.

bangerth on 9 May 2018

@bangerth My final set of pull-requests is #6549, #6548, #6547. I am happy to tag rc4 immediately after. The testsuite should be fixed now (see updated readme on gmsh requirements).

tamiko on 9 May 2018

Tell me what to do with #6548 and we'll do the rc4 and then call it a day.

bangerth on 9 May 2018

OK, merged.

Now go with rc4 before anyone changes their mind :-) (I know that there are still the AD doc patches out there, and also the update for step-18. I also appreciate the work that @jppelteret and @drwells have put into this, but I do think that we need to come to a conclusion at one point and we seem to just be piling up things as we go, without coming to convergence.)

bangerth on 9 May 2018

👍1

@bangerth

(I know that there are still the AD doc patches out there, and also the update for step-18. I also appreciate the work that @jppelteret and @drwells have put into this, but I do think that we need to come to a conclusion at one point and we seem to just be piling up things as we go, without coming to convergence.)

i thought the conclusion was to at least add AD documentation https://github.com/dealii/dealii/pull/6542 which @jppelteret promised and delivered.
Let's not rush out the release just because of 1-2 days difference.
I am sure if there would be no goal of getting it in 9.0, @jppelteret would invest time to do it right now (as he's overloaded with work these days) and I certainly would not try to help reviewing it at 12 at night.

davydden on 9 May 2018

It would also be worth to consider merging a fix for #6552 (introduced by #6536), or reverting #6536.

tamiko on 9 May 2018

...the update for step-18. I also appreciate the work that @jppelteret and @drwells have put into this, but I do think that we need to come to a conclusion...

For #6473 I am going to submit a patch reformatting the table of images today. Our discussion ended with us deciding not to make any code or result modifications.

jppelteret on 9 May 2018

@davydden and @jppelteret -- fair enough. I do appreciate the work that both of you are putting into this!

bangerth on 9 May 2018

@bangerth Let me put it that way - if the three remaining PRs/issues linked above is all that is left, we had a pretty successful release marathon. (With a bit of last-minute fallout, but that's why we have release candidates and bug-fix rounds in the first place.) :-)

tamiko on 9 May 2018

@dealii/dealii btw, i think we shall have another release task -- update deal.II in package managers (Spack, Candi)

p.s. similar to

17 Ask Luca for Mac packages

davydden on 9 May 2018

The release task list is in the release-papers git repo -- feel free to add entries there!

bangerth on 9 May 2018

The release task list is in the release-papers git repo -- feel free to add entries there!

done. Updated in the post above as well.

davydden on 9 May 2018

@tamiko @luca-heltai create_tria_02.debug still fails for me even after I added tetgen and netgen to gmsh. @tamiko with what external libraries do you build gmsh with?

davydden on 9 May 2018

@davydden , are you using the gmsh installation that comes with spack? That's what I used.

luca-heltai on 9 May 2018

I have

$ gmsh -info
Version          : 3.0.6
License          : GNU General Public License
Build OS         : Linux64
Build date       : 20171105
Build host       : gmsh.info
Build options    : 64Bit Ann Bamg Bfgs Blas(Generic) Blossom C++11 Chaco DIntegration Dlopen Fltk Gmm Jpeg(Fltk) Kbipack Lapack(Generic) LinuxJoystick MathEx Med Mesh Metis Mmg3d Mpeg Netgen ONELAB ONELABMetamodel OpenCASCADE OpenGL OptHom PETSc Parser Plugins Png(Fltk) Post SLEPc Solver Taucs TetGen/BR Tetgen1.5 Voro3D Zlib
FLTK version     : 1.3.4
PETSc version    : 3.7.5 (complex arithmtic)
OCC version      : 7.1.0
MED version      : 3.2.0
Packaged by      : gitlab-runner
Web site         : http://gmsh.info
Mailing list     : [email protected]

masterleinad on 9 May 2018

for me it is:

~
691-heltai@turbolo: spack[develop] $ gmsh -info
Version : 3.0.6
License : GNU General Public License
Build OS : MacOSX
Build date : 20180505
Build host : turbolo
Build options : 64Bit Ann Bamg Bfgs Blas(Custom) Blossom C++11 Chaco DIntegration Dlopen GMP Gmm Kbipack Lapack(Custom) MPI MathEx Mesh Mmg3d NativeFileChooser Netgen ONELAB ONELABMetamodel OptHom Parser Plugins Post Solver TetGen/BR Tetgen1.5 Voro3D
Packaged by : heltai
Web site : http://gmsh.info
Mailing list : [email protected]
~

luca-heltai on 9 May 2018

are you using the gmsh installation that comes with spack? That's what I used.

yes, including those additions https://github.com/spack/spack/pull/8060 I was failing for me prior to those changes as well...

$ gmsh -info
Version          : 3.0.6
License          : GNU General Public License
Build OS         : Linux64
Build date       : 20180509
Build host       : davyddenubuntu
Build options    : 64Bit Ann Bamg Bfgs Blas(Custom) Blossom C++11 Chaco DIntegration Dlopen GMP Gmm Kbipack Lapack(Custom) LinuxJoystick MPI MathEx Mesh Mmg3d NativeFileChooser Netgen ONELAB ONELABMetamodel OptHom Parser Plugins Post Solver TetGen/BR Tetgen1.5 Voro3D
Packaged by      : davydden
Web site         : http://gmsh.info
Mailing list     : [email protected]

btw, it works on macOS but it fails on Linux!

davydden on 9 May 2018

@tamiko @luca-heltai create_tria_02.debug still fails for me even
after I added tetgen and netgen to gmsh. @tamiko with what
external libraries do you build gmsh with?

Version          : 3.0.6
License          : GNU General Public License
Build OS         : Linux64
Build date       : 20180508
Build host       : kestrel
Build options    : 64Bit Ann Bamg Bfgs Blas Blossom C++11 Cairo Chaco DIntegration Dlopen Fltk GMP Gmm Jpeg Kbipack Lapack LinuxJoystick MathEx Mesh Mmg3d Mpeg NativeFileChooser Netgen ONELAB ONELABMetamodel OpenCASCADE OpenGL OptHom Parser Plugins Png Post Solver TetGen/BR Tetgen1.5 Voro3D Zlib
FLTK version     : 1.3.3
OCC version      : 7.2.0
Packaged by      : portage
Web site         : http://gmsh.info
Mailing list     : [email protected]

In combination with opencascade-7.2.0

tamiko on 9 May 2018

this test fails on half of the testers https://cdash.kyomu.43-1.org/testSummary.php?project=1&name=gmsh%2Fcreate_tria_02.debug&date=2018-05-09

davydden on 9 May 2018

@davydden On tester with updated gmsh it works fine.

tamiko on 9 May 2018

On tester with updated gmsh it works fine.

I also have Blas, netgen, tetgen enabled, but this does not help. Whateve “updated” means, it’s not just that.

I build dealii with the latest OCE, if I recall correctly, but gmsh isn’t build against it

davydden on 9 May 2018

I build dealii with the latest OCE, if I recall correctly, but gmsh isn’t build against it

You probably also need gmsh being built against OCE. That's the one
difference I can see right now.

tamiko on 9 May 2018

@davydden Any luck to convince gmsh to pass the test? :-)

tamiko on 10 May 2018

@masterleinad @davydden @Rombur Give me one more testsuite spin on current dealii-9.0 head, namely v9.0.0-rc5-6-ge549f8f0bf, please :-)

tamiko on 10 May 2018

What did you decide about #6542?

bangerth on 10 May 2018

@bangerth The consensus so far is to cherry-pick #6542 to the release branch once it is merged - which is about now :-)

tamiko on 10 May 2018

OK, we should then be set now!

bangerth on 10 May 2018

@luca-heltai @kronbichler @jppelteret Please have a critical look at https://tamiko.kyomu.43-1.org/tmp/doc/

tamiko on 10 May 2018

Thanks @tamiko! I picked up a few small typo's in the documentation I recently wrote. See #6564.

jppelteret on 10 May 2018

@tamiko I've checked my recent additions and everything looks as it should. Thanks!

kronbichler on 10 May 2018

@tamiko

Any luck to convince gmsh to pass the test? :-)

looks like gmsh build with oce did the trick, I am rerunning the tester now, give it 4-5 hours to complete (it's a standard oldish desktop PC).

davydden on 10 May 2018

OK, let's do two more small documentation updates #6570, #6571.
The last remaining question that I would like to discuss is, what to do with #6567.

tamiko on 10 May 2018

I'd say #6567 falls in the category of "nuisance" and not "critical" at this point in a release cycle.

bangerth on 10 May 2018

Regarding our code gallery:

Pull request to fix "ceres": https://github.com/dealii/code-gallery/pull/45
We get a couple of deprecation warnings: https://github.com/dealii/code-gallery/issues

Apart from this, all examples configure, compile and run fine. (Only checked for functionality, not correctness of the computational result.)

tamiko on 10 May 2018

@bangerth, le me disagree on #6567. Independently on how we tackle the issue that @code @endcode is used both for inlined code and for example code, I think the indentation issue is worth addressing anyway.