Daniel Kahn Gillmor <dkg@fifthhorseman.net> writes:
> It makes no sense to stop indexing a message just because one of the
> parts failed to decrypt, so i'm not going to immediately return.
>
> I'm willing to accept that only the last log message will make it out to
> the caller, and i could track whether anything has been written to the
> log and change the return value in that case. would that be acceptable?
>
That sounds like an improvement. Other options I can think of
- accumulate an error string. With talloc_asprintf_append, this is
not _too_ terrible. Making a second logging function [1] that
didn't clear the log buffer but appended would maybe make sense
(aside from contradicting what I said in the previous message).
This would still need some status return to alert the caller.
- Pass a logging callback; this requires API changes. We
already have a such a callback for notmuch_database_compact.
While thinking about this, I noticed several suspect uses of
_notmuch_database_log in current index.cc, at least in
_index_mime_part. These are probably my fault, resulting from
blindly replacing printfs.
[1]: untested:
void
_notmuch_database_log_append (notmuch_database_t *notmuch,
const char *format,
...)
{
va_list va_args;
va_start (va_args, format);
if (notmuch->status_string)
notmuch->status_string = talloc_vasprintf_append (notmuch->status_string, format, va_args)
else
notmuch->status_string = talloc_vasprintf (notmuch, format, va_args);
va_end (va_args);
}