Re: [Netatalk-devel] [Netatalk-admins] noadouble

[Anton S. <ant...@gm...>]

On Dec 29, 2009, at 5:00 PM, Frank Lahm wrote:
>>>> It creates resource fork when user explicitly want to store there some non-default information.
>>> 
>>> ...this is exactly the way "noadouble" is expected to work afaict.
>> 
>> But it isn't work in that way. try to mount options:noadouble volume and SetFileInfo on some file on it.
>> It will fail. (OK it will not fail, but it will not create resource fork and store flags).
> 
> Yes, I got that point. I just wanted to give you a clear indication on
> the expected behaviour.

Yes :) It is behavior I've been expecting according to manual, but reality always cruel! :)

>>>> And there is some other issue, I have to check still, but I got impression that you sometimes open files with
>>>> O_RDWR when it isn't really unnecessary (O_RDONLY can be enough).
>>>> Am I wrong?
>>> 
>>> Yes. Afair we need O_RDWR for share modes. The real details might be
>>> more complicated, but that's basically it.
>> 
>> But then, de facto, you change mtime all the time. Which isn't good. Am I right?
> 
> Ah, wait, I guess I got you wrong. As it seems you were referring to
> the "normal" data fork, were you? I thought that you were referring to
> the adouble files. It's them to which the semantics I described
> relate.

Yep, I refer to data-fork. But I still have to check it, to be sure.
But my point is that changing mtime without real need isn't simplify life for backup system :)

Anton.

PS: Dammit! It looks like netatalk is in rapid development. I wrote my  symlink patch for 3-days old HEAD, and most of it rejected with current HEAD! :)