Is there a way to exclude .git/COMMIT_EDITMSG files from file-name-history? Now that I'm authoring commits with Magit, those commit message files are often the most recent entries in file-name-history, which is annoying since they're never what I'm looking for when scrolling through history entries.
These files (COMMIT_EDITMSG, git-rebase-todo and others) are added to file-name-history by server-visit-files. It is possible to undo that by advising that function, and I have implemented that.
But I am unsure whether that advise should also remove entries for these files that already existed prior to that function being called. I think so because those entries are very likely there from before this advice was added. On the other hand this might surprise users who, for whatever reason, edit these files by manually opening them. I'll decide at a later time.
file-name-historyand notrecentf-list? If it is the latter, then you can userecentf-exclude. On the other handgit-commit.elalready includes(eval-after-load 'recentf '(add-to-list 'recentf-exclude git-commit-filename-regexp)), so you shouldn't have to. – tarsius Sep 20 '16 at 17:10find-fileand such should be added tofile-name-history(I think). Do you do that? Why? – tarsius Sep 20 '16 at 17:12find-fileon aCOMMIT_EDITMSGfile, but when I usefind-fileand hitM-p(previous-history-element), there's usually aCOMMIT_EDITMSGin there. To confirm, I first deleted it from file-name-history, inspecting the variable to ensure it was gone. Then I used Magit to author a commit, and inspecting file-name-history again showed it back in there. – ivan Sep 20 '16 at 18:14file-name-historyare wrong. – tarsius Sep 20 '16 at 18:38server-visit-filesadds the file tofile-name-history. I don't see any way to prevent that short of advising that function. – tarsius Sep 20 '16 at 19:01