[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-core-devel
Subject:    Re: Upcoming 2.2 release schedule, bickering and bugs
From:       Dawit Alemayehu <adawit () kde ! org>
Date:       2001-06-21 12:38:55
[Download RAW message or body]

On Thursday 21 June 2001 02:11, George Staikos wrote:
[snip]

>   - kio_http and kio_ftp (along with ldap and nntp which are not much of a
> concern right now) are not using TCPSlaveBase.  This means they are
> duplicating a lot of code, and quite honestly, some of the code in kio_http
> is outdated.  I don't plan to update the SSL code there.  Porting to
> TCPSlaveBase will fix a fair amount of this.

Though this would make both yours and my life much simpler, it is IMHO something
that can wait for 2.3 release as we've discussed before.  Porting kio_http to TCPSlaveBase
is something that can be accomplished relatively easily given enough time.  However, ensuring
that it works properly and does not break anything else is completely a different matter.  AS
such rather than wasting time on this, since it is going to be re-written anyways, I rather we
hold off until 2.3.  This will give us enough time and testing cycles to work out any kinks that
might exist as a result of the re-write.

> In addition, kio_http is becoming very unmaintainable very quickly.  We've seen
> many bugs pop up recently and code get broken due to the spaghetti nature that
> is evolving.

Well, I am definitely not going to argue with you on the hard to maintain spaghetti nature
of the code, because it is.  However, I'm just curious about what are the "many bugs" you
were referring to ?  Are these related to SSL ?  Otherwise, there are very few issues (hint:
countable with one hand) problems with kio_http itself. /me confused by this statement...

> Proxying may still need more work.  Dawit can provide more feedback on
> this, but I'm betting that we still don't support SSL proxies very well.

Well I'm almost there on this.  Infact, I am finishing up the config dialog stuff as
I write this which should complete all the proxy related problems with the exception
of SSL tunneling.  Even that only fails if the very first connection is a  request for
a secure document through an "authenticating proxy".  Otherwise,  everthing 
works just fine.  So I think we can safely cross this issue out as a show-stopper
for 2.2 :)

> This is just KIO alone.   This should be enough evidence that 2.2 needs
> more work.  I don't think any of these are really new features.  They're
> cleanups and bugfixes by my definition.  (and of course licence fixes)
> Again...  We don't have an urgent reason to get 2.2 out the door right
> away. Let's hold back and clean things up some more.  I'm sure the docs and
> translation teams won't mind either.  I know that portions of KDE are
> undocumented (either for users or developers) so it would even be a time
> for others to catch up too.  I'm not promoting adding any new features. 
> Just cleaning up what we have... tying up loose ends...  finishing
> incomplete code.
>
> I really don't want to see a big 2.2.1 patch shortly after 2.2.0 fixing
> all the bugs that we could have fixed 2.2.0 but couldn't because we rushed
> it out the door.  I think this would also give us a chance to sort out kwin
> issues.  I don't know much about this code, but judging by the mailing list
> traffic, I think a few problems have surfaced there.  Let's give the kwin
> coders some time to reorganise and put together a plan to deal with this. 
> I imagine there are many little things that could be fixed in khtml and
> it's components too.  Need I continue?
>
>  I say again....  Let's slow things down, think over our plans for 2.2
> and what we wish it to be, and fix the things that really need to be fixed.
>  2.2 will be the third major release (2.0,2.1,2.2) in less than a year. 
> That's a very impressive release schedule.  I don't think anyone can
> complain if it's closer to the end of that first year.

I would have disagreed with this, a first for me since I'm always the one to object
to what I feel are rushed releases:),  had it not been for the removal of the SASL
code today which seems to have broken features that were largely implemented due
to high demand from end-users.  I think this along with getting some of the things that
were messed up as a result of the recent fallout should be enough to at least force
us to re-evaluate whether we need to revise the current release schedule.

Just my $0.02 worth.

Regards,
Dawit A.

[prev in list] [next in list] [prev in thread] [next in thread]