New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update boost to version 1.64 + minor enhancements #194
Conversation
We normally don't install all the headers availabe in boost. There is an interesting set of functionalities like multiprecision and odeint that while we are not using, but would be interesting to have available. This is done to ease future boost updates but also in the hopes of making it possible to use this functionality in a near future. No functional change expected.
Modern processors should handle this without any trouble.
Sorry for the PR mess. |
Other than the updates and bug fixes, this version works really well with older compilers and systems and is the last one to include TR1: which we use as a last resort for really old systems lacking standard C++ compliance. https://www.boost.org/users/history/version_1_64_0.html
Builds on Windows with 1.55. Building module salEntering /cygdrive/c/Source/openoffice/main/sal/inc Entering /cygdrive/c/Source/openoffice/main/sal/osl/w32 Making: all_cpposl.dpslo Entering /cygdrive/c/Source/openoffice/main/sal/textenc Making: all_textenc.dpslo Entering /cygdrive/c/Source/openoffice/main/sal/rtl/source Making: all_cpprtl.dpslo Entering /cygdrive/c/Source/openoffice/main/sal/systools/win32/uwinapi Entering /cygdrive/c/Source/openoffice/main/sal/systools/win32/onlineupdate Entering /cygdrive/c/Source/openoffice/main/sal/osl/all Compiling: sal/osl/all/debugbase.cxx 1 module(s): Reason(s): ERROR: error 65280 occurred while making /cygdrive/c/Source/openoffice/main/sal/osl/all When you have fixed the errors in that module you can resume the build by running:
|
The latest version should work. |
Bummer! This version of boost was tested was tested on our supported compiler upstream so it might be some local build issue. |
I will try to do that. |
Looking at your d.lst changes, there is a dll and respective /detail directory which may be needed for windows. Those didn't exist in 1.55 |
I will try that! |
This comment was marked as off-topic.
This comment was marked as off-topic.
Builds fine on Windows now.
@jimjag Can you please try to build on macOS/Linux? |
@Pilot-Pirx Thanks so much for finishing the bits for Windows. Since it's getting so long to merge(!), let me provide some motivation: We currently make limited use of boost, so the boost update will provide some fixes and performance enhancements, but perhaps more important is that it provides functionality that modern languages would regularly provide and that are currently unavailable in our older C++ environment. |
Hi Pedro, I am highly motivated to merge this PR but I didn't want to do it without being sure that it builds on macOS and Linux. Since it builds on BSD how high is the probability that it also builds on macOS? |
Very high .. FreeBSD and macOS share most of libc and FreeBSD adopted natively LLVM as well, but there are sufficient differences that one can't be 100% sure. The one thing that gives us relative confidence is that upstream tested all the OS/compiler combinations we are interested in: Linux:
OS X:
Windows:
FreeBSD:
|
Thanks for the info! I can merge it, although it still is a "shot in the dark" for me. For Linux I can see the results from our buildbot (trunk): If successful, I would cherry-pick for AOO42X. @jimjag What do you think? |
Can you cherrypick the 1.55 updates and squash the 1.64 update in a different commit? |
Yes, I can "Squash and merge" it to trunk. If the Linux buildbot is successful, I will cherry-pick it to AOO42X tomorrow. |
Buildbot does not build: 1 module(s): Reason(s): ERROR: error 65280 occurred while making /home/buildslave/slave/openoffice-linux64-nightly/build/main/sal/rtl/source When you have fixed the errors in that module you can resume the build by running: build --from sal |
Not being a developer I will now step back and let others take over... Feel free to reopen this PR. |
I wish the 1.55 changes had been merged separately. The sinc_pi change was useful despite the version. In any case it appears that the boost build per-se didn't fail but that the built left missing headers unistalled that are required by the linux port. I can't really follow up at this time as I don't have a linux box available to test. |
Unfortunately GitHub does not seem to allow to re-open a PR once it was merged... If you don't mind (and my time permits) I would like to do "remakes" of your original PR. |
Add some headers to the existing 1.55 boost internal package.
Make cleaner usage of boost's sinc_pi functionality. There is no need to limit the policy.