@bdarcus Thanks for that link, I did come across that as well. It bothered me quite a bit that these instructions were very old and clearly written for DT 2.0 which was new like, in the previous millennium or something. I figured that all those dependencies were superseded by generations of versions since that time. I don’t want to end up with a darktable which is too different from the version being built by the DT team and their CI server.
Perhaps I’m not expressing myself clearly enough here. I’m not looking for a 3.0 release build, I want to try to build Darktable from source so I can test out new fixes and features as they are committed.
Some distribution X may very well have a repo for DT 3.0 but that doesn’t help me very much in my mission here.
If you want to continue to use the docker container, you need to allow the container to write data to the host system using volumes: Use volumes | Docker Documentation
I wondered about that, but just tried it, and it worked. The dependency list doesn’t specify specific versions, so you end up with the latest versions.
Being able to build from source imho is a major requirement for foss. RT devs spent a lot of time to make it easy to build RT from source. Would be good if building dt from source also would be that easy. Tough I have to admit, that I did not try to build dt from source…
@rgo Thanks, this looks very interesting. Much more up to date information. It probably doesn’t help me produce a statically linked version of darktable but it’s very nice to have updated build information, so thanks for that.
Just to follow up I did manage to run the Darktable version I built inside docker, where all the dependencies are properly installed.
The problem with running GUI applications inside Docker is that the X server has to be shared with the container. It’s a bit of a pain to get that off the ground, but in case anyone else wants to give this a try there is a handy tool called x11docker available on Github, which - with the proper options, is perfectly able to run Darktable from inside a container, with zero performance loss.