Re: [evolution-patches] camel-uid-cache.c robustness patch

[Not Zed <notzed ximian com>]

On Thu, 2003-06-05 at 06:07, Jeffrey Stedfast wrote:
> Here's another go at it, removed the mkdir_hier() code and #include'd
> camel-store.h for that implementation (so we don't have to have 2 copies
> of it).

We still do :)  ... there's another one in camel-file-util (apart from
the one in e-util i think).

But that was me, i thought having to include store.h in thigns that
weren't a store, for a totally unrelated utility method wasn't very
clean.  But i never got around to propagating it to other callees.

I guess the rest of it looks ok.  I'm not sure of the real utility of
the try-to-save-as-much-as-possible logic; its pretty complicated for
what it gives you.  And you're probably much more likely to hit serious
issues with appending to the mailbox before then, which would generally
cause this to fail anyway.

> Also took care of all the issues (except the g_strsplit() issue) that
> were brought up in the previous patch.
> 
> You'll note that a failed write() won't necessarily mean we won't still
> overwrite the previous cache with the new cache. there's some (hopefully
> correct) logic to figure out if we should try to overwrite the old cache
> anyway in an attempt to save as much as we can.
> 
> There's a biggish comment in the code explaining the logic so I won't
> repeat it in my mail.
> 
> Jeff