Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Jan 30 01:42
    luz-arreola commented #409
  • Jan 30 00:53
    luz-arreola commented #409
  • Jan 30 00:51
    luz-arreola commented #409
  • Jan 29 17:43
    stefanseefeld commented #409
  • Jan 29 16:13
    luz-arreola opened #409
  • Jan 16 01:53
    stephengtuggy closed #405
  • Jan 16 01:53
    stephengtuggy commented #405
  • Jan 10 19:30
    sdarwin commented #398
  • Jan 09 00:24
    stephengtuggy commented #405
  • Jan 06 09:36
    mont29 opened #408
  • Jan 06 07:02
    Osyotr commented #405
  • Jan 06 00:53
    stephengtuggy commented #405
  • Dec 24 2022 13:10
    Osyotr commented #405
  • Dec 20 2022 14:44
    owillebo commented #400
  • Dec 14 2022 15:23

    github-actions[bot] on gh-pages

    deploy: 47d5bc76f69e20625214381… (compare)

  • Dec 14 2022 15:17

    mclow on boost-1.81.0

    (compare)

  • Dec 13 2022 12:43
    huangqinjin synchronize #406
  • Dec 13 2022 09:07
    hdu-sdlzx commented #407
  • Dec 13 2022 09:06
    hdu-sdlzx edited #407
  • Dec 13 2022 09:05
    hdu-sdlzx edited #407
Stefan Seefeld
@stefanseefeld
Here you can see how faber is tested on Appveyor: https://ci.appveyor.com/project/stefanseefeld/faber/build/job/2f4wvrai22l05a2d. At some point I run faber --info=tools cxx and it reports all MSVC versions it can see: https://ci.appveyor.com/project/stefanseefeld/faber/build/job/2f4wvrai22l05a2d#L40
I assume if you run faber --info=tools cxx you get an empty list, right ?
Hakim
@h4k1m0u
It's weird, both faber --debug and faber --info return the same previous message. I don't think the message comes from another program, since where faber points to the script file in Python's directory
Stefan Seefeld
@stefanseefeld
oh, so you are saying that it's the execution of faber itself that triggers the error ? I assumed it was one of the commands run by faber.
In that case (and just to confirm this is the problem), can you run faber by using python3 path/to/faber/script instead ?
Hakim
@h4k1m0u
That seems to work, it's building boost_python
Stefan Seefeld
@stefanseefeld
(note that the tests actually run python3 scripts/faber .... the python3 setup.py install execution should actually generate a faber.bat script that's then run.
Do you mind submitting a bug report to https://github.com/stefanseefeld/faber/issues ?
Hakim
@h4k1m0u
Sure
Stefan Seefeld
@stefanseefeld
Many thanks !
sahaj jain
@ssaahhaajj
@ssaahhaajj
hey is the project of python library for gsoc still open to select anyone?
because i think it is easy for me to work in it
and if yes i can send you the implementation of python's set in c++ asap
Stefan Seefeld
@stefanseefeld
just replied to you in the private channel
7vikpeculiar
@7vikpeculiar
@stefanseefeld
I am interested in contributing to boost-python, any pointers on how to get going?
NIRMAL MANOJ C
@NirmalManoj
Hi everyone, I'm Nirmal, a CS undergraduate student from IIIT Hyderabad, India. I would like to work with boostc++ for this year's GSoC. I want to connect with this community to know more about this org. But, many of the boostorg's chat rooms are not accessible. So can someone please tell me the chat room I should join be joining in case this is not the right one? Asking because this seems like a sub-room for detailed discussions.
klakho0400
@klakho0400
hey, I would like to contribute to boost.python how can i get started and which group should I join
vipbuoy
@vipbuoy
hey, i want to ask how we work here?
Rohan-Zanjal
@Rohan-Zanjal
Can anyone help me how can i start contributing
Tom Kent
@teeks99
So it sounds like travis-ci.org is going to get a lot more restricted for open source projects. Any thoughts on migrating python's CI? Github Actions?
Stefan Seefeld
@stefanseefeld
Hi @teeks99 , thanks for your help ! I'm trying to familiarise myself with github actions now. It seems this could even cover other platforms such as Windows, so we could get rid not only of travis-ci but also appveyor
Tom Kent
@teeks99
Yeah, they have linux, windows, and mac...so all the major platforms.
What do you think of the pull request that does the 8 linux configurati
ons? I was thinking it would be best to get the easiest ones going first.
Stefan Seefeld
@stefanseefeld
Yeah, I think I'm going to merge them. Still trying to understand all the details...:-)
Tom Kent
@teeks99
Anything I can do to help explain it?
Stefan Seefeld
@stefanseefeld
Is there anything we need to do to enable github actions ? How can I test your PR in the target repo ?
Stefan Seefeld
@stefanseefeld
Another (more specific) question is: looking at https://github.com/boostorg/python/pull/335/files#diff-1a8cf21570cf96acf7408fd6572a61a36aae89cb22dd3563de3b936f9e89bcb8 I wonder how you install faber. Is this already included in your docker image ?
Tom Kent
@teeks99
Yes, both the gcc and clang image get faber, as well as the BCP'ed part of the tree for python dependencies.
I don't believe there is anything you need to do to enable them, but I could be wrong. I didn't have to do anything for my forked repo. Testing the PR is more difficult. Might be easiest to bring that one file in on your own branch in the boostorg/python repo. This, being an infrastructure change, is a bit different than a normal code change.
Stefan Seefeld
@stefanseefeld
OK, thanks for the explanation. I'm experimenting right now with setting this up for the faber project. I'm almost done (Ubuntu and Windows builds are working, still looking into some OSX issues, then comes doc generation).
As soon as all this is working, I'll move on to Boost.Python. I expect to complete this before the end of the year.
Stefan Seefeld
@stefanseefeld
@teeks99 , do you have experience with OSX ? I have set up a workflow (for faber only, for now) for Ubuntu, Windows, and OSX, but the latter produces link errors: https://github.com/stefanseefeld/faber/runs/1583570864?check_suite_focus=true
I wonder whether I missed anything in the Python setup...
Stefan Seefeld
@stefanseefeld
Hmm, I start to wonder: I thought only on Windows do python extensions need to be linked against libpython. The above error looks like on OSX, too.
Tom Kent
@teeks99
I'm definitely not a mac expert. Before my try that failed yesterday morning (see the comment in the github issue) it had probably been a decade since I last worked with a Mac. I think the problem I was having is something similar.
Stefan Seefeld
@stefanseefeld
I think I understand the issue: -undefined,dynamic_lookup is missing somewhere
Stefan Seefeld
@stefanseefeld
Faber migration to github actions is complete (https://github.com/stefanseefeld/faber/actions), now on to Boost.Python. :-)
Tom Kent
@teeks99
Sounds good!
Stefan Seefeld
@stefanseefeld
I have a few comments on boostorg/python#335, and as we are here, perhaps it's easier to do this live...
Can you squash all commits into one, but leave out the change to the appveyor config ? That way the PR will be a single commit, adding a single file. Also, as you said yourself, the 1.66 release is rather old, so perhaps this would be an opportunity to move to a more recent boost release as baseline ? (only if this is easy and quick to do)
Otherwise this looks great, and I'm really thankful for your help with this.
(I would actually like to clone this workflow as soon as it's committed to run OSX and Windows builds. Once that's done, we can decommission travis-ci as well as appveyor.)
Stefan Seefeld
@stefanseefeld
@teeks99 (^)
Tom Kent
@teeks99
Is there not an option in the pull request to squash the commits? If not, you may need to turn that on for the repo.
I wanted to get things working first, then I'll put out another pull request to move to 1.75. Also, I'm still having a small issue getting 1.75 to BCP correctly in my docker image, so I want to get to the bottom of that.
OSX and Windows will look a lot different. I don't believe there is any way to run docker images with OSX, so we'll need to do all of that setup logic in boost::python's workflow. As I mentioned in the boostorg/python#334 issue, I've got a branch started with Mac OSX support, but I'm having serious problems with it. You want to have a look and see if you have any ideas?
Tom Kent
@teeks99
For windows, docker images are supported, but the environment isn't quite as settled as docker on linux. I do have a VS2019 docker image (src), but it isn't quite ready for prime time...and I don't have any older versions dockerized yet. I'd vote that we stick with the windows image from github for now. Plus Appveyor is still working fairly well, so this isn't quite as urgent.
On that note, the appveyor scripts are tied to exact version numbers of visual studio...which was breaking when the appveyor people updated their image. Is there a way to only specify the major version of visual studio (or the msvc toolset, i.e. 14.0, 14.1, 14.2)? I'm not familiar with fabers toolset lookup options.
There is a very simple pull request for appveyor waiting out there that simply rolls the VS2019 version from 16.8.1 to 16.8.3....but that's a losing battle. We'd need to update every few weeks.
Stefan Seefeld
@stefanseefeld
Hi Tom, sorry I just now saw your messages. Yes, there is an interface at github for squashing. But I really want to exclude the change to the .travis.yml file, to keep things separate. So if you can remove that commit from the PR, we can merge it (with squashing). Moving to 1.75 in a second PR is perfectly fine. There will be a couple of iterations until we get the full coverage (including OSX) that travis-ci now provides.