On 22-05-2025 11:59 pm, Jeff King wrote: > On Thu, May 22, 2025 at 05:27:15PM +0000, Aditya Garg wrote: > >> Upon setting up imap-send config file, I encountered the very first bug. >> An error showing "no imap store specified" was being displayed on the >> terminal. Upon investigating further, in static int git_imap_config, >> cfg->folder was being incorrectly set to NULL in case imap.user, imap.pass, >> imap.tunnel and imap.authmethod were defined, and the values that these configs >> intended to set were not being set at all. > > I read "these configs[...]were not being set at all" as imap.user, etc. > But I think the only thing affected was imap.folder, which was > incorrectly being reset when we saw the other fields (and of course the > leak-fix for those fields was not kicking in correctly). > > So: > > [imap] > host = example.com > user = foo > folder = INBOX > > was fine, but: > > [imap] > host = example.com > folder = INBOX > user = foo > > was not (we end up with a NULL folder variable). > >> Because of this, git imap-send was basically not usable at all. The >> bug seems to be there for quite a while, and has not yet been >> detected, likely due to better options like git send-email being >> available. > > I think that probably explains why it was not detected (by users or the > tests). It was dependent on the usage and ordering of particular config > options. > > (The patch is still doing the right thing, of course; I'm just trying to > add more context to the commit message). mhm, looks like I need to read the code again.
