ESC meeting minutes: 2020-02-27

[Jan-Marek Glogowski]

Am 27.02.20 um 16:54 schrieb Miklos Vajna:
> * GSoC 2020: offer meson build system transition as an idea? (Jan-Marek)
>    + I’m interested in mentoring this.
>        - anybody else interested as a 2nd mentor?
>    + there is private/jmux/meson, based on https://github.com/jpakkane/core
>        - just my playground for a day to get a feeling for meson
>            - had some minor problems because of missing cmake error logging
>              (got help in #mesonbuild and from jpakkane)
>            - separate target for install / instdir
>            - didn’t yet check alias target for gbuild module emulation
>            - not sure, if partial builds work, ignoring depends?!
>                - add a function to include/ header
>                - just build some modules, not all dependencies
>                - maybe just a use case I’m used to rely on for VCL changes
>        - added instdir as install target and VCL qt5 + kf5 builds
>        - currently builds ~80 of ~200 LO libraries
>          (includes offapi and officecfg generation)
>        - contains a few pre-generated headers like sfx slots
>          (~ 95k of 120k lines overall change are pre-generated headers)
>        - needs externals as system libraries
>          (I used an Ubuntu focal schroot; can share setup, if wanted)
>        - always builds out-of-tree
>        - working with it “feels” a lot like high-level gbuild
>    + I won’t add it to ideas, if ESC doesn’t think it’s worth a slot
>    + want to evaluate other build systems instead? Not my GSoC idea.
> 
>    + question: do we want to offer a slot for the meson topic?
>      + need a mentor → Jan-Marek could do that
>      + need a secondary mentor
>      + make it explicit that this is a test on a feature branch (Stephan)
>        + a feasibility study
>        + since there is no consensus that we want to switch
>      + issue: student does the project → we need to maintain it (Stephan)
>      + we have people around who understand gbuild, e.g. Michael S,
> Lubos (Lubos)
>      + the risk is: the student builds a lot of custom stuff on top of
> meson (Ilmari)
>      + a dozen import filters are built using autotools on Windows with
> gcc-wrapper (Michael S)
> 
> => with the above limits, no objections (all)

So I've added
https://wiki.documentfoundation.org/Development/GSoC/Ideas#Feasibility_study:_building_LibreOffice_using_meson

Jussi Pakkanen, the original meson upstream author, offered to co-mentor
this. But getting a 2nd LO developer (as a 3rd) in addition would be
nice to have too.

Jan-Marek