[PATCH] Support for POSIX ACL in kioslave/file and kio/kio

[Till Adam]

On Sunday 07 August 2005 23:05, Thiago Macieira wrote:
> Till Adam wrote:
> >Heya folks,
> >
> >please find attached two patches, one against kdelibs/kioslave/file and
> > one against kdelibs/kio, which together implement support for managing
> > POSIX ACLs on filesystems that support it. Please review, I'll then
> > merge this into 3.5.
>
> Thanks, Till, it's great work.

Thanks.

> >Limitations/TODO:
> > o it currently only works for a single file (no support for "partial"
> >   handling), that's the next thing I'd like to tackle once this is in
>
> Do you plan on having an interface to set the ACLs on "this dir, all files
> and dirs beneath it"?

That already works. Simply check the "reply recursively" checkbox when editing 
the properties of a dir. :)

> > o the users and groups are offered as comboboxes, which has scalability
> >   issues, I'm planing to offer a lineedit with completion when the
> > number of available users/groups exceeds a certain limit
>
> I recommend here a standard widget for selecting users and groups, like
> kaddressbook integration in KMail's composer -- in fact, kabc could
> integrate with it too -- but outside libkabc.
>
> This kind of widget in libkdeui could be useful for other applications as
> well as the standard properties dialog (think chgrp).

Good idea.

> > o the configure checks aren't done yet, I guess this needs to be a
> > compile time option
>
> I'll help you here. But I'll need you to tell me how the KACL class will
> behave when the support for it isn't present. It must exist and be BC to
> the working version.
>
> The same goes for kacl.h: it should be installed regardless of the support
> being actually compiled in.

Ok.

> >I'd be grateful for any testing and feedback that you might be able to
> > spare.
>
> I don't have any filesystems with ACL support, so I can't help you
> actually test the code, but I can at least make sure it won't break what
> existed before :-)

That's just as important, thanks.

> Why does the copy constructor for KACL parse the rhs's string
> representation? Can't it copy the parsed data?

True. I'll fix that.

> Don't we have a standard class for representing and fetching system users
> (I'm thinking of KACL::getUserName and KACL::getGroupName)? If we don't,
> we should.

I didn't find anything, but I agree that it might be a good idea. How does 
this tie in with the recent discussions on core-devel about "disply of system 
users"? Is there code to that effect somewhere already? We should probably 
provide something like that for KDE 4, at least.

> BTW, thanks for fixing the setuid/setgid/vtx flags in kio_file. I had
> never noticed them to be wrong, not even when I fixed the same thing in
> kio_ftp

I think credit for that goes to Gyoergy, I took that from this patch, I 
believe.

Thanks for reviewing,

Till
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <https://mail.kde.org/mailman/private/kfm-devel/attachments/20050807/81bdbb78/attachment.sig>