Created attachment 711804 [details] build.log.gz Hi! pms-7_p20180430-r1 build continues indefinitely on pdflatex eapi-cheasheet — I killed it after 40 minutes of CPU time on Core i7-4790 (it is used to build 32-bit packages for another host). This is likely a bug in pdflatex, so I'm assigning to its maintainer. pdflatex is from texlive-core-2021, it is usable and I can build another documents with it.
Created attachment 711807 [details] emerge --info
What is in the build.log looks good to me, and shows pdfTeX running just once (it is supposed to run in total 5 times). I suppose you've killed it at that point? It shouldn't take 40 minutes for that first pass, so something is definitely going wrong.
(In reply to Ulrich Müller from comment #2) > What is in the build.log looks good to me, and shows pdfTeX running just > once (it is supposed to run in total 5 times). > > I suppose you've killed it at that point? It shouldn't take 40 minutes for > that first pass, so something is definitely going wrong. Yes. But that happened during quite complicated @world update. After the following loop terminated on no-op on both actions: emerge @preserved-rebuild emerge --depclean I can no longer reproduce this problem and pms builds fine. I suppose something in texlive was broken during @world upgrade, because the loop above involved some texlive packages.
FWIW is/was probably a real bug here. I have been hitting (and ignoring) this problem on multiple machines for ages and finally discovered (today!) that the solution is to forcibly uninstall dev-tex/leaflet (no longer in the tree) in order to successfully update app-doc/pms. This was a dependency of prior versions of app-doc/pms and depclean won't remove it (unless using --with-bdeps=n) until app-doc/pms is upgraded, which won't happen automatically with because the build hangs forever with leaflet installed.