Re: patchfs, audio update

["Andrew V. Samoilov" <sav bcs zp ua>]

Pavel Roskin wrote:
> Hello!
>
>
> > Well, but some quoting (and security) issues fixed in the CVS just now.
> > Patch attached. BTW, new incarnation reports wrong patch sizes for
> > gzipped patches.
>
>
> Just for your information.
>
> Size is not very critical after I disabled fast loading on non-local VFS
> in the editor.  You can open files with 0 length, e.g. INFO/DESCRIPTION in
> rpm archives, and it will show properly.  I'm not saying that reporting
> the right size is not important, but it's not as important as it was.

Negative file size is illegal for vfs parser.  Such entries are skipped.

> I haven't analyzed the patchfs code, but if finding the file size needs a
> lot of additional work (e.g. uncompressing the patch one more time), it's
> better to report "0" and hope that "copyout" will do the right thing
> nevertheless.

copyout is ok.

Adam Byrtek 'alpha' wrote:
> On Wed, Dec 18, 2002 at 06:45:42PM -0500, Pavel Roskin wrote:
>
>>Sorry, I don't understand why an empty line is not a diff boundary. 
>> It's not valid in diff, as far as I know.
>
>
> Yes, it is not a valid diff.
> Valid diff puts:
> ' \n' (space and a newline)
> But Andrew gave me a link to a www mailing list archive, with a patch
> inside <pre> tags. When one saves such webpage (or paste&copy it) the
> space is stripped and it looks like this:
> '\n' (only newline)
>
> I think we should treat this as a diff contents even if it isn't
> formaly valid.

Well, I resent this letter to Adam, it's valid diff.
Problems starts when you compress this patch by gzip or bzip2.
It seems perl's tell() works incorrect for pipes (I wonder), as far as 
lseek() (I don't wonder).


>>Besides, Andrew has CVS access and uses it very responsibly.
>
>
> Maybe, but his fix isn't necessary - at least for my system. See:
> http://www.rpg-portal.pl/alpha/mc-patchfs.png
>
> Lets wait for Andrew's reaction if you don't trust me.

We trust you, but sometimes you misunderstand us.

>
>>I think that "@@\n$" should be replaced with "@@" in both places.
>>Tested, works for me.  I'm not applying it to avoid conflicts, but 
keep it
>>in mind.

Good stuff.  Also I want to replace all '^-{3}' with more readable '^---'.


--
Regards,
Andrew V. Samoilov