TexLive 2017 has been released over a month ago now, and I expect the update to be nontrivial. I thought I'd create this issue to track the updating process.
cc @vcunat
Yes, highly nontrivial, probably. I don't think it will get to the top of my TODO within a few weeks at least.
Would also like to highlight the lack of a repstopdf symlink, as mentioned here https://github.com/NixOS/nixpkgs/issues/4210
Eh, well, I'll try adding that, though I really dislike their tendency to create a hundred executables. epstopdf --restricted will do exactly the same...
@NickHu, @7c6f434c: this patch seems ~good~ usable 7aa140488aa – can you test repstopdf in your use cases, to be surer?
It adds repstopdf link unconditionally even if epstopdf doesn't exist?
I am not sure that hacking a repstopdf package with a single symlink into texlive package set is worse (I just install it in addition to some combined TeXLive expression)
@7c6f434c: it only adds repstopdf conditionally on having epstopdf. I also tested that. (cp would fail anyway)
Ah nice. If only I remembered what document required repstopdf to test…
@7c6f434c: your thesis? :-D https://github.com/NixOS/nixpkgs/issues/4210#issuecomment-179920807
I had a hope there is a less hopeless test case, because in case of my thesis there is also disser package and bibliography style compatibility…
Hm, on the other hand, it breaks on eps include _first_, so I can still test even if I do not actually succeed.
@vcunat erm, in what branch is that repstopdf fix?
@7c6f434c, it's in my fork: https://github.com/vcunat/nixpkgs/commits/texlive-repstopdf
It might be easiest to simply run curl https://github.com/NixOS/nixpkgs/commit/7aa140488aa.patch | git am in such cases...
One could naively expect that if GitHub agrees to show it in NixOS/NixPkgs, there is a branch of this repository that contains the commit. I certainly did.
It includes some "related" repositories, apparently, and I misused that here. It would be nice if they showed some reachability information, though. EDIT: it might actually be a simple side-effect of their optimizing all forks together...
It would be even nicer if Git handled branches in a saner way…
Looks like this hack does allow my thesis compilation to fail on bibliography instead of repstopdf.
Any news on texlive 2017 ?
@Nadrieril: not on my side, at least.
@NickHu, @7c6f434c: this patch seems good usable 7aa1404 – can you test
repstopdfin your use cases, to be surer?
The repstopdf created this way worked on our end. However the check for epstopdf being a symlink seems to fail.
Most helpful comment
Hm, on the other hand, it breaks on
epsinclude _first_, so I can still test even if I do not actually succeed.