Closed seisman closed 4 years ago
Today is the scheduled date for GMT 6.1.1 release.
To me, these two issues may need to be fixed before 6.1.1 release. They're not critical issues, so I'm also OK to leave them to 6.2.0 release.
@joa-quim is looking at #4003. I will see what the other one is about.
Cannot look at it before tomorrow night
Sent from my iDedo
No dia 22/08/2020, às 01:43, Paul Wessel notifications@github.com escreveu:
@joa-quim is looking at #4003. I will see what the other one is about.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.
All the tests (except a few known failures) pass on Linux, macOS and Windows. PyGMT also works well with 6.1 branch.
The MB-System really is set up to work well with homebrew. Would it be possible for @seisman to just clone that repo and run /configure and see if all is found. I am running into various proj stuff "missing"
Just cloned the repo and run ./configure
. It can find my GMT 6.1, but can't find openGL:
Did not find OpenGL headers anywhere, use --with-opengl-include to set the location
configure: error: Did not find OpenGL headers anywhere, use --with-opengl-include to set the location
Might have to add these to configure:
--with-opengl-lib=/System/Library/Frameworks/OpenGL.framework/Versions/A/Libraries --with-opengl-include=/System/Library/Frameworks/OpenGL.framework/Versions/A/Headers
or there is also a switch to turn off the building of the interactive stuff that requires all that.
If you are cloning master it may be related to development of the new Qt interface (just aguess).
Yes, I cloned the master branch. I have /System/Library/Frameworks/OpenGL.framework/Versions/A/Libraries
but don't have /System/Library/Frameworks/OpenGL.framework/Versions/A/Headers
.
Fortunately, I have X11 installed, and X11 provides the GL headers and library. So I'm running
./configure --with-opengl-include=/opt/X11/include --with-opengl-lib=/opt/X11/lib/
make
It's still compiling. So far so good. Will let you know when it finishes.
GMTSAR builds fine with 6.1.1. I have no tests for it.
Building of MB-system looks good to me. I don't see any errors or warnings, and I don't run it.
I think that is fine. Caress told me they do CI tests with GMT now so hopefully we would be told - but likely more of a master 6.2 issue.
I ran gallery through 27, but 28 fussed Download cache file: @Kilauea.utm.nc+Uk but for 6.1.1 release I think it is generally fine. I am fixing a few outdated syntaxes in gmt_gallery.m as well. Many changes to our core examples means some are failing because cache files are no longer used.
Checklist has edit changelog but I cannot find that file anywhere?
Checklist has edit changelog but I cannot find that file anywhere?
It's doc/rst/source/changes.rst
.
Thanks, perhaps revise the checklist to state that. Since 6.1.1 has no new features, and I dont really want to hunt down all those entries, should we just add a blanket statement such as "the GMT 6.1.1 release adds no new features but fixes a number of bugs that have surfaced since the release of 6.1. As such, it is a recommended upgrade for all 6.1 users". ?
Yes, I agree. No need to list the bug-fixes, as there are too many.
a relatively large number of bugs
I always get these two when building the docs - master as well - so nothing new. I assume this means Add-h is undefined in two places where the explain-h.rst_ file is included?
I do not think there is any changes needed in INSTALL.md since it is at least major version insensitive, no?
I assume we/you can check the GMT.jl box for 6.1.1, @joa-quim ?
I always get these two when building the docs - master as well - so nothing new. I assume this means Add-h is undefined in two places where the explain-h.rst_ file is included?
See the fix in #4148. Our CI should warn us if there are warnings when building the documentation. Will see why it doesn't work later.
I do not think there is any changes needed in INSTALL.md since it is at least major version insensitive, no?
Yes.
I will build the tarball and bundle candidates, assuming @joa-quim has no issues after testing Julia.
Julia tests are fine. The dinner hopefully as well.
Files placed on my soest ftp release directory as listed in admin/build-release.sh
28b4f59190fe401f9d94511bcc8e08546c6e8104e51eed806dcf1164d62582db gmt-6.1.1-darwin-x86_64.dmg
9706ffa858a98bb1b97162e3887aa03c6c746017a274951e94a6341a59efad12 gmt-6.1.1-src.tar.gz
d476cba999340648146ef53ab4a3f64858cbd2f5511cdec9f7f06f3fb7896625 gmt-6.1.1-src.tar.xz
Time to build the Windows installers plus test the src and bundle. I will be on road for a few hours but will continue later.
Sorry, being copied...now all 3 are there.
I can see these files, but can't download them (it says these files don't exist).
Sorry, default permissions. Will fix in 1-2 hours...
We are jumping again the RC step.
Do we need that for bug fix release you think? 6.2 yrs, but 6.1.x?
I don't see any difference between minor and point releases regarding the possibility of having screwed something. GDAL & PROJ (the two I follow) don't make any distinction.
And as I mentioned before, an RC should automatically promote to final if no issues are found.
Permissions fixed.
I don't mind doing RC etc . We would need to update the checklist probably. I was thinking since the only stuff we place in 6.1 are corrections, the likelihood of finding problems are low. yes, there are still bugs of course but there always will be and they are in master too.
Might @seisman know a slick way to set file permissions on a remote server? Seems it cannot be done via scp, so may have to run chmod on the remote server. Perhaps
ssh pwessel@ftp.soest.hawaii.edu 'chmod og+r /export/ftp1/ftp/pub/pwessel/release/*'
Yes, that seems to work and will add to the script.
The macOS bundle and source tarballs look good to me.
The 6.1.0 release miraculously escaped a breaking change done on the eve because of tests. That type of things (and we always do last minute changes) happens equally in all type of releases. So, if we do RCs we should do it for all releases.
Tested on a clean Win. Posting in two sites because I'm curious if the ualg site is faster for you.
http://fct-gmt.ualg.pt/mirone/gmt-6.1.1-win64.exe http://fct-gmt.ualg.pt/mirone/gmt-6.1.1-win32.exe
http://joa-quim.pt/mirone/gmt-6.1.1-win64.exe http://joa-quim.pt/mirone/gmt-6.1.1-win32.exe
@joa-quim Have you tested the ghostscript transparency issue (#3798)?
Yes, I did sometime ago (like I said in that issue). What I did now was to compile 9.53 version but stripped out of the annoying warnings of future deprecations.
Great! The Windows installers look good.
Will soon have all files on ftp server. Question: I assume I make the 6.1.1 tag while branch 6.1 is the active branch?
I assume I make the 6.1.1 tag while branch 6.1 is the active branch?
YES.
OK, I have done my parts, I guess the CI will build the GitHub release next.
OK, I have done my parts, I guess the CI will build the GitHub release next.
Yes, but the bad news is that the CI fails (https://github.com/GenericMappingTools/gmt/runs/1063178680) because it can't download files from the FTP. Permission issue again?
Yes... I changed the permissions after placing all the files in the pub.release but just coping them to the gmt dir reverted to defaults... OK, please tell it to run again.
The CI is done. You can review the draft release now.
Looks OK to me - not sure what to review other than reasonable file sizes etc.?
and the checksums are good.
Looks OK to me - not sure what to review other than reasonable file sizes etc.?
and the checksums are good.
I believe that's enough.
Will you take care of the draft edit and links, @seisman, and I can start making an announcement message?
Version: 6.1.1
Scheduled date: Sep 1, 2020
Before release:
src/gmt_make_*.sh
to update some .c and .h filesadmin/gs_check.sh
to test if latest ghostscript version workscmake/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
set (GMT_PUBLIC_RELEASE TRUE)
line3rd-party update