You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Maybe I'm just TOO used to having M-x shell be a text buffer. But afaict this works properly in mac's terminal.app as well. If I run something and then search for text I expect all of them to be highlighted. In ghostel only the fully matching strings w/o wraparound are highlighted.
The problem is that to solve that cleanly without hacks would require a bigger re-write to not write hard newlines that libghostty gives us, but let Emacs do the wrapping. This comes with a few negatives as well.
PR #608 implements a kludge for isearch, occur and evil-search. But so far nobody had this problem so
if you do NOT use isearch, maybe I'm not even merging it as it fixes only a part of an edge case (searching over line-wrapped lines).
I use isearch. I get that there's a conflict with "the emacs way" and "the libghostty way"... this might just be out of scope... but these kludges you've added are just that: kludges. They're evidence of the conflict.
I figure I should keep reporting the friction I'm experiencing. Some will get kludges. Some with get closed. But maaaaybe at some point we'll find some middle ground where there's a unification of the kludge over the whole conflict that allows for all the regular tools to start to work a little better.
I'm kinda wanting some sort of (with-ghostel-buffer &rest body) macro that would for the duration make the buffer an "emacs buffer" (rather than a libghostty owned/responsible buffer) and temporarily restore emacs' notion of a buffer, soft line wrapping, etc... That would make it easier to advise-wrap things like isearch, occur, and all the other nonsense I've been filing issues on. :P
Activity
dakra commentedon Aug 7, 2026
How do you search? isearch?
The problem is that to solve that cleanly without hacks would require a bigger re-write to not write hard newlines that libghostty gives us, but let Emacs do the wrapping. This comes with a few negatives as well.
PR #608 implements a kludge for isearch, occur and evil-search. But so far nobody had this problem so
if you do NOT use isearch, maybe I'm not even merging it as it fixes only a part of an edge case (searching over line-wrapped lines).
dakra commentedon Aug 11, 2026
@zenspider ping
zenspider commentedon Aug 12, 2026
I use isearch. I get that there's a conflict with "the emacs way" and "the libghostty way"... this might just be out of scope... but these kludges you've added are just that: kludges. They're evidence of the conflict.
I figure I should keep reporting the friction I'm experiencing. Some will get kludges. Some with get closed. But maaaaybe at some point we'll find some middle ground where there's a unification of the kludge over the whole conflict that allows for all the regular tools to start to work a little better.
zenspider commentedon Aug 18, 2026
I'm kinda wanting some sort of
(with-ghostel-buffer &rest body)macro that would for the duration make the buffer an "emacs buffer" (rather than a libghostty owned/responsible buffer) and temporarily restore emacs' notion of a buffer, soft line wrapping, etc... That would make it easier to advise-wrap things like isearch, occur, and all the other nonsense I've been filing issues on. :P