Re: [PATCH] lib: make find_message{,by_filename) report errors

Subject: Re: [PATCH] lib: make find_message{,by_filename) report errors

Date: Tue, 4 Oct 2011 08:00:46 +0300

To: David Bremner

Cc: Notmuch Mailing List, Austin Clements

From: Ali Polatel


David Bremner yazmış:
>On Sat,  1 Oct 2011 11:12:23 +0300, Ali Polatel <polatel@gmail.com> wrote:
>> From: Ali Polatel <alip@exherbo.org>
>>
>> Looks like the patch did not make it correctly the first time.
>> Resending using git-send-email™
>>
>> You may also find the commit in my notmuch repository:
>> git://github.com/alip/notmuch.git branch: find_message
>
>
>Hi Ali;
>
>Thanks for reworking this patch. I looked at branch find_message-v2
>in your repo. I have a few comments.

Thanks for going over the patch, expect a new set of patches soon!

>- In the comments for _resolve_message_id_to_thread_id I guess thread_id
>  should be thread_id_ret?

Fixed.

>- in notmuch_database_find_message_by_file_name, I'm not sure why you
>  set status to NOTMUCH_STATUS_OUT_OF_MEMORY in the catch block. Is this
>  a typo?

Looks like a copy & paste error. I must have blindly copied the error
from the previous block. Fixed.

>- after the DONE: label of the same routine, how is *message_ret destroyed?
>  does it need to wait until the talloc context "notmuch" is freed?

Yes, I have modified it to call notmuch_message_destroy() in case
'*message_ret' is non-NULL after the DONE:

>- I don't really get the change of user to caller around notmuch.h:286
>  It is not a big deal, but I guess we should try to be consistent.

I don't get what you mean by consistency here but this hunk is unrelated
to the problem which the patch is trying to address.
Reverted.

>David
>
>

         -alip
part-000.sig (application/pgp-signature)

Thread: