I’m confused (yeah, so what else is new ?): there are 4 ‘collections’ of source files to use when building dt. Given the warning in the announcement on the mailing list (“As always, please don’t use the autogenerated tarball provided by github, but only our tar.xz file.”) and given that I am unable to identify which is the ‘autogenerated tarball provided by github’, I would appreciate some clarification of which is the correct set of source files to use.
Collections 1 and 2 are identical, which is re-assuring. Collection 3 has 904 fewer files than either 1 or 2, which makes it look suspicious. Collection 4 has 1368 more files than either 1 or 2, even after ignoring files in the /build/ folder or those in the /.git/ folder. Some of these look like they are rather important - those to do with Rawspeed for instance.
My (probably defective) logic says build from the source that is downloaded by the ‘git clone’ command. In which case why not say so on the darktable.org web-site instead of exhorting (in the announcement letter) the use of ‘collection’ 1, above?
If you simply clone darktable, you end up on the master branch. You need a git checkout to switch to whatever branch you need.
Cloning and getting all submodules probably gives you the integration tests, which you don’t really need. Those can be excluded; search the forum, or wait until I get home.
For just building a working version of darktable, use your 1. :
the source package provided by the devs. That is the source most users will grab, and so any possible problems with them surface quickly. Option 2 should be the same file as option 1…
Git (option 4) is useful if you want to use a development version or are hunting a bug. If you only want to use stable versions, it’s more hassle than it is worth.
Note that if you use a provided package (options 1, 2, 3), GIT commands are not needed or desired, and probably just won’t work.
Btw, I have a suspicion that the release-4.6.1.tar.gz (option 3) is the tarball provided by github, which you are told explicitly not to use
You have not understood my intent: I am very well aware of the admonishments - and their location - to not use the source tarball provided by git. As I tried to convey, the issue is knowing where/what this tarball is - which would best be defined by name, for somebody such as myself, not familiar with the terminology or the significance of suffices like '.gz; and '.xz;.
Even a supported on this forum, like’ revietor’, with orders of magnitude more knowledge and understanding then me, ‘has a suspicion that … is the tarball’ - implying that the tarball identification is not clear.
So the ‘better wording’ I would suggest would be to state the fully qualified name of the file that is to be avoided, or some other way of uniquely identifying it.
Oh, that was because you gave the same link for both, just the filenames were different (and I didn’t want to check which is which).
It’s (for me) perfectly clear which version to get when I go there… and you don’t have to go to github to get the correct tarball, there is a direct link to the correct file on the darktable site itself (where most users would start anyway).
There is no reason to use github at all if you just want the source for the latest published version, e.g. because there’s no compiled version available for a particular distribution (yet).
For me, Github is basically to get a development version, or to do bug hunting, or development (where you need to work against the latest sources). (or, of course, if there are no source packages provided… not the case for darktable). Using git just to get the latest stable version is overkill.