kdelibs/knewstuff

[Andras Mantia]

On Thursday 03 March 2005 23:51, Josef Spillner wrote:
> Am Donnerstag, 3. März 2005 22:00 schrieb Andras Mantia:
> > Don't fail when trying to upload two resources one after the other.
> > The meta data file name was not re-generated. There is still a
> > problem that the error dialog appears and quickly disappears, but I
> > don't know the reason of it.
>
> What did you try to do - uploading two files which should belong
> together or one upload and later another one?

One upload after the other. 

> In the first case, archives should be used and in the second, it
> should work without problems. Is this a specific upload case in
> Quanta? Because I didn't see such problems yet.

It may be specific to Quanta as I don't know how others use the 
KNewStuff classes. :-) For example try to upload two toolbar resource 
in Quanta one after the other. The second will silently fail, or to be 
more exact, the data file is uploaded, but the meta is not. What Quanta 
might do differently is:
- it uses KNewStuffSecure (but here I don't see a problem, as it simply 
calls at the end KNewStuff::upload(m_signedFileName, QString::null) and 
this is where the problems begin)
- it keeps the KNewStuffSecure objects in the memory, so it doesn't 
destroy and create them each time. I think this is what triggers the 
problem.

>
> I don't think the library supports such parallelisms, for instance
> there's a race condition if the provider file is requested twice at
> once. 

There is no parallelism involved here. Just that if after an upload is 
finished and you request a new upload using the same KNewStuff object, 
the locally created meta file will have the same name, even if you 
change the Name field, and the upload fails.

> Also, if the file is not yet processed on the server, an upload 
> of the same meta file will always fail (this is about one minute).
> Actually each following upload of the same file should be rejected
> because it is in the database already, at least the revision number
> should be increased by the author.

The problem here is that the local KNewStuff client creates a metafile 
that is already in the server, altough IMO it should not.
I agree that it should reject the upload if it's already on the server, 
just that I have no idea why the error dialog disappears without having 
the chance to read it. Did you see this bug?

Andras


-- 
Quanta Plus developer - http://quanta.sourceforge.net
K Desktop Environment - http://www.kde.org
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://mail.kde.org/pipermail/kde-core-devel/attachments/20050304/4f042b1e/attachment.sig>