On 2024/08/21 23:32, José Maldonado wrote: > You are right about the use of wordexp. In fact, the proposed patch is > actually just a rollback to the old string_to_path function that dunst > used in its 1.9.2 version (current port version) and that works fine > for our case (we can't use things like $HOME in the config or it gives > an error reading the path for scripts).
Ah - that wasn't obvious - in which case I'd be happy enough with that if a comment is addded to the top of the patch file like 'revert to before commit XXX which changed this to use wordexp'. > For now, if we want to do the bump version, the rollback is a "good > option", until we do a code equivalent to the current wordexp using > glob. > > In any case, the best way to solve this is to make a PR that allows us > to correct this situation so that it doesn't hinder the portability of > dunst in OpenBSD for future versions. I will get to work on it and if > it is accepted, it is likely that the next version will not need this > dirty hack to compile and work. If sending something upstream for this, please either just replace the wordexp code completely, or gate it on a test for whether wordexp is available rather than detecting OpenBSD.