Closed PaulWessel closed 8 months ago
When we update/release the new remote datasets?
Yes, we should be able to release the datasets the same day.
We should probably add a tickbox for "Updating all new remove data to main oceania server".
I suggest we aim for 6.5 release on Dec 1, otherwise we'll not get there in 2023. I know there are outstanding issues with bugs but there always are. No way to fix everything unfortunately. We can always to a 6.5.1 if there are good reasons after fixing some bugs. @maxrjones, might you be able to help with the ChangeLog again?
@maxrjones, might you be able to help with the ChangeLog again?
Yes, no problem. do you anticipate any noteworthy changes between now and Dec 1?
Want to update the gsfml repo but trouble. Wonder is one of our CMake experts can help?
@maxrjones, might you be able to help with the ChangeLog again?
Yes, no problem. do you anticipate any noteworthy changes between now and Dec 1?
Sorry the delay... Since Dec 1 is tomorrow I think so. Maybe wait to do anything until I post again here.
I suggest we aim for 6.5 release on Dec 1, otherwise we'll not get there in 2023. I know there are outstanding issues with bugs but there always are. No way to fix everything unfortunately. We can always to a 6.5.1 if there are good reasons after fixing some bugs.
No worries if people aren't available in Dec such that this isn't a possibility, but I wanted to float the idea of a Dec 15 release date so that it's after AGU and there's no chance of accidental regressions causing issues for presenters before the conference.
Yes, good idea. In the olden days we often released closer to AGU and caused panic!
I don't think the AGU argument makes much sense anymore (and people only update GMT if they want/know-how). I would rather get it out before 8 December.
It is true that now we have close to 1200 tests while in the olden day we would make code changes, have no tests, and then gotten bitten by various issues, often on different platforms. Of course, nobody is forced to update software a few days before AGU. I know many who wait weeks and months and even years to update their macOS version, so it is always buyer be aware (caveat emptor).
I agree that AGU is no longer an argument. Also, with all the package installers, reverting becomes trivial. Let's set our own release dates that are convenient for all.
We should just pick a date and stick to it. We have yet to release a GMT version that did not have known bugs the we did not get fixed in time. We cannot wait for a miracle that fixes all our bugs, so better to release a 6.5 that is superior to 6.4 so that those who need official releases can benefit.
I suggest Friday Dec. 8. Is there any of you that are able to build the macOS bundle for Intel?
We could start building it today no?
I am thinking of adding the GSFML (Global seafloor fabric and magnetic lineations) as an optional supplement (or maybe just another supplement) since people are having trouble making it from Cmake, such as ME!... I will make a branch to see if this is a 1-day job or longer.
I also would like to merge the shakemap
branch. Not used since I wrote it but it worked back than and I hate to loose good(?) code. Branch has docs and builds fine. No tests though.
Hi @maxrjones, no more new things will be added to GMT after both @joa-quim (grdvs30 and grdshake added to seis supplement) and I (added gsfml supplement) have done our thing. Safe to do ChangeLog now.
Hi @maxrjones, no more new things will be added to GMT after both @joa-quim (grdvs30 and grdshake added to seis supplement) and I (added gsfml supplement) have done our thing. Safe to do ChangeLog now.
Sounds good, I'll be able to do this later today. BTW I no longer have an intel mac.
Right, and I have one in Hawaii I so far have been unable to update macports on... Does anyone else on this post have access to an Intel mac?
I do. And I can do the macports Portfile update, unless @seisman gets there first.
@PaulWessel I am assuming we're not going to include all the longoptions updates in the changelog and will just have announce them all once complete - is this correct?
No, no word about long-options until GMT 7 probably.
BTW, after much yelling and cursing I finally got mac port to install GDAL... SO looks like I can build the installers but I will ask several people to try so they know what to do.
Oddest thing yesterday. Built the macOS bundle on M1 Silicon and when I tried to used it it immediately crashed! Rebooted, updated MacPorts with any changes there the last few days, and voila, worked again. Also got my macnut in Hawaii updated to the latest macOS Monterey the old gnome (2014) can handle, Xcode etc and now can build the Intel bundle as well. I suggest we simply pick Friday Dec 15th. As of today:
If I have forgotten something critical here (not listed in checklist) please ping asap. Yes, there are many more bugs listed in the issues but it is not possible to fix everything in a finite short time. There is always 6.5.1 if something stupid shows up.
I just clicked "freeze the code" since we are drawing a line in the sand today. Only allowed to fix typos in docs and add to changes.rst.
Nice to see 100% of tests pass again.
4. Hopefully @federico is back by then and he and I can (for the first time) test the "duplicate the candidate server to the oceania server". I am thinking of first duplicating Oceania on the unused test server in case we learn some nasty lessons. But candidate works well so don't expect troubl
I am back. How should I try this?
Do git pull on gmtadmin-server
for a few tiny updates tonight. You should able to test-run the eventual update of oceania. If you do make test-release
the script will build a script, copy it to the server /tmp
, and run it. This will place everything from candidate to test server. You may wipe the test server first via make test-delete
. If you want to be very cautious then comment the last lime in the scripts/test-release.sh
file so it does not actually run the script it copies to the server. Examine the script it makes manually to see if it looks reasonable - ask me if you have questions or see any monkey-business.
At some point you can probably undo that comment and start from scratch and run make test-release
again. Then you can do export GMT_DATA_SERVER=test
in a terminal and then try to make some new plots. Make sure it creates the right directories in ~/.gmt/test/...
OK?
Sorry, just added some fixes I forgot to commit.
I am testing it. It seems that it is working fine. But I am getting some errors because of this dir (/export/gmtserver/gmt/test/server/earth/earth_relief2.5
). @PaulWessel could you delete it please.
PaulWessel the test server
is working fine. I test it with remote_map_check.sh (of the remote-dataset repository) including earth_mdt and I got all the maps.
Wonderful! Thanks Federico, we are ready for the server release once we get 6.5 released first. A few days maybe.
Struggling with my ancient Mac Pro Intel [2014] again. When running the full build I get this odd message:
ninja: error: loading 'build.ninja': No such file or directory
The CmakeError.log says
ld: Library not found for -lSystem
Also finding errors related to "missing" include files io.h and direct.h which are Windows files. WTF? As it is I am unable to build the x64 bundle. Anyone else who can?
I have placed the two tar versions and the arm64 for Mac installer in my pub/releases directory:
faac08659811eaa2e5336e654d62ba0d3ecd2168ac9f7157f3da5a69bba7ddae gmt-6.5.0-darwin-arm64.dmg e7af80ec3e5735fd9536b53b2b920b9f96a435b59d63421f61e21fe2a11960a2 gmt-6.5.0-src.tar.gz 75198c690d2babf54cea8adb47697b0dcf23af219da3dac52eda9812c3a75478 gmt-6.5.0-src.tar.xz
in ftp.soest.hawaii.edu:/export/ftp1/ftp/pub/pwessel/release. Hopefully, @Esteban82 and @anbj can test those on Linux and maybe @seisman and @maxrjones can check the M1 bundle?
@maxrjones, might you be able to look at the ChangeLog again since pre-AGU and make any updates to changes.rst? Down the road it may be useful to have your recipe/steps in writing in case others have to step in and do that task.
Paul, is this the URL? https://ftp.soest.hawaii.edu/pwessel/release/ What is that string? is the checksum?
You'll have to experiment I think since ftp is tricker than it used to be. Play with the extra directory pub as well. Yes the strings are the checksums (sha256 hash) as it says in the build-release.sh file.
Built and tested gmt-6.5.0-src.tar.gz. Looks good.
I'll be happy to test the Windows portable package.
Should we merge #8038 first?
See #8254
Win installer and portable
https://fct-gmt.ualg.pt/tmp/GMT5/gmt-6.5.0-win64.exe 0B4D47058E0B00F20A467DBA260AAB6238B705E8C6C6FA5B0D7FB237410532E2
https://fct-gmt.ualg.pt/tmp/GMT5/gmt-6.5.0-win64.zip DE7E05F87896FFB87E73F24EFB2FE21BE608CDED02AE0C3569652866E1448658
Great, I have placed them in the SOEST ftp release with the others. Might you add the sha256 tags here?
Edited.
Tested a couple of commands on the portable version, and looks very good.
@maxrjones, might you be able to look at the ChangeLog again since pre-AGU and make any updates to changes.rst? Down the road it may be useful to have your recipe/steps in writing in case others have to step in and do that task.
Looking now - very sorry for dropping the ball on this!
I have rebuilt the two tar source balls and the macOS arm64 installer. Here are the latest sha256 hashes:
5c0bf3c24de9c806500eec8d5900b6869a7698bfc468b4edc0155bfffbfdcefe gmt-6.5.0-darwin-arm64.dmg
294b1b8a724329beab9790d62e29aca7fbdedba3fb21de6b3f662b1a80a52fdf gmt-6.5.0-darwin-x86_64.dmg
b17e165fd6c85aeb0a281700bd89522af8c2676a2d7bdb51a6b242fa9f1779c9 gmt-6.5.0-src.tar.gz
4022adb44033f9c1d5a4d275b69506449e4d486efe2218313f3ff7a6c6c3141e gmt-6.5.0-src.tar.xz
I am unable to build macOS x86-64 installer on my Mac Pro sitting in Hawaii as the signing fails. Is there anyone who have an Intel mac that can help build the installer, i.e., cd to top gmt dir and run admin/build-release.sh with MacPorts?
I am unable to build macOS x86-64 installer on my Mac Pro sitting in Hawaii as the signing fails. Is there anyone who have an Intel mac that can help build the installer, i.e., cd to top gmt dir and run admin/build-release.sh with MacPorts?
I've done my best. Does this work for you?
Sorry, that one is not proper, as I still have not been able to get the example images from dvc
OK. I take it you have installed dvc but have you gotten authorised (?) via @maxrjones to interact with DVC. I guess it should not be necessary to pull from DVC. So what is the trouble if you are at the top dir (which contains subdirectories cmake, src, etc) and run dvc pull
?
Version: 6.5
Scheduled date: Jan, 5, 2024
Before release:
src/gmt_make_*.sh
to update some .c and .h filesadmin/gs_check.sh
to test if latest ghostscript version worksdoc/rst/_static/version_switch.js
if it's a minor releasecmake/ConfigDefault.cmake
GMT_VERSION_YEAR
is current yearGMT_PACKAGE_VERSION_*
is correctly setGMT_LIB_SOVERSION
is correctly setGMT_PUBLIC_RELEASE
toTRUE
GMT_VERSION_DOI
Release:
After release:
GMT_PACKAGE_VERSION_*
incmake/ConfigDefault.cmake
#8263set (GMT_PUBLIC_RELEASE TRUE)
line #82633rd-party update
Volunteers needed! Please let us know if you volunteer to help to maintain GMT in these 3rd-party tools.