Re: [Netatalk-devel] CNID refactored.

[Rafal L. <rle...@pr...>]

Hello, 

> It doesn't compile if
> DB_VERSION_MAJOR > 4 || (DB_VERSION_MAJOR == 4 && DB_VERSION_MINOR >= 1) 
> 
> is true
Yes, I've also found it may not compile if file mangling is disabled.
Will be fixed soon. Seems like this is also broken in HEAD.
If you already have patches, send them :)

> And there's a mess with indentation.
I'll fix it tommorow. Propably will be a massive change so send me
patches rather than commit into CVS until this will be done.

> Anymay db3 doesn't work, (error in open), db3 was in bad state in HEAD. 
> so maybe use cdb by default as --with-cnid-default-backend=xx doesn't 
> work yet.
Will be investigated tomorrow. Or if you already have patches, send them
to me :)

> return cdb line is missing in cnid_(cdb|db3)_new, I guess it works by 
> luck, eg cdb is in eax.
Oops ! Fixed.

> I missing something or there's a 'typo' in directory.c ?
> @@ -251,8 +248,6 @@
>       /* cname is not efficient */
>       if (cname( vol, ret, &ptr ) == NULL )
>           return NULL;
> -#endif
> -    return dirsearch(vol, did);
> ^ Huh?
> }
> T2_spectest catch it, ie it dumps a core.
Fixed. Looks like one 'dd' keystroke to much :(

> How do you want to work? I'll send you patch or I'll commit to the branch?
Sending patches may be better until I'll get rid of
deeper overhauls (like, cleaning up indentation).
Then I'll let know (soon, I hope).

Thanks for review. As usual, some severe bugs have been spotted ;)


-- 
Rafal Lewczuk <rle...@pr...>
"Pronet" Nikołajuk Rozmus Muliński Spółka jawna