Add a missing parameter in advice on font-lock-fontify-keywords-region

[cpitclaudel]

This breaks various other modes in recent Emacs, most noticeably git-region-history:

Error during redisplay: (jit-lock-function 1501) signaled (wrong-number-of-arguments #[(ad--addoit-function beg end loudly) "\306� \203 \0\307\310\311
!\312\313Q!!\203 \0\310\311
!\312\313Q!�
#\210��
#\211�)\207" [ad-return-value proof-buffer-type proof-assistant-symbol beg end loudly nil fboundp intern symbol-name "-" "font-lock-fontify-syntactically-region" ad--addoit-function] 6] 3)

[erikmd]

Hi @cpitclaudel, thanks for your PR for master!
Did you test your patch with async as well?

[erikmd]

Otherwise I can first merge #238 in async (which significantly improves the behavior w.r.t. retracting), and afterwards we test your patch on the new async so obtained?

[erikmd]

BTW could you precise which recent Emacs version you tested, and which mode contains the function git-region-history ?

[cpitclaudel]

I tested the current master, but things have been broken for a while. The font-lock-fontify-keywords-region has taken the third argument since at least 1995, and our advice was just wrong.
I'm happy with any merge strategy (I don't expect this to conflict)

[erikmd]

OK, so thanks for having spotted this longstanding bug!

Just for the record, could you give more details about git-region-history? is it related to Magit?

[cpitclaudel]

Just for the record, could you give more details about git-region-history? is it related to Magit?

No, not at all :) It's called by vc-region-history as vc-git-region-history-mode, which is a function vanilla Emacs.

[erikmd]

OK :) I regularly use Emacs' Git facilities (with commands having C-x v prefix in vanilla Emacs) but wasn't aware of M-x vc-region-history, thanks for the tip!

[erikmd]

Successfully tested after merging locally this PR in latest pg: async. Will merge in both branches.