On Tue, Mar 04, 2025 at 03:27:38PM +0100, Rene Kita wrote: > On Tue, Mar 04, 2025 at 02:19:00PM +0100, Theo Buehler wrote: > > > Unfortunately I'm not an Emacs user and don't have time to dig into > > > this. If there is anything to do from the w3m side please make sure keep > > > me Cc'ed. > > > > It is definitely a problem in w3m, not directly OpenBSD related. It's > > another instance of https://todo.sr.ht/~rkta/w3m/59 > > > > As mentioned by Jeremie and Stuart, the problem involves w3m -dump_source > > and w3m -dump_extra because the proc function pointer is used with > > uninitialized garbage and we end up calling loadSomething with it: > [...] > > The workaround I committed is to initialize it to NULL: > > > > https://cvsweb.openbsd.org/cgi-bin/cvsweb/ports/www/w3m/patches/patch-file_c?annotate=1.8 > > This is the same fix which was the reason for the mentioned ticket #59 > on my tracker. This results in segfaults on other sites as reported.
Indeed. > I reverted the original commit because of the report as I couldn't > reproduce the issue I tried to fix with commit e933cf83. Now that I can > reproduce I will try to find a proper fix. that would be nice, thanks. > Aside, when applying patches to a port, shouldn't the ports package > version reflect the patch level? If I'm not mistaken version in -current > is still w3m-0.5.3pl20241116p0 where I would have expected a p1. A port starts out with no revision: w3m-0.5.3pl20241116. I set the REVISION=0 when adding the patch: w3m-0.5.3pl20241116p0, then comes p1, etc.
