[Framework-Team] Re: PLIP #215: Include new KSS versions

[Andreas Zeidler]

On Jan 30, 2008, at 9:33 AM, Balazs Ree wrote:
> Hi Andi,

hi balazs,

> my mail is not coming up on the freamwork list, which I don't  
> understand....
> anyway here is the cc.

i'll cc the list and try to quote all of your mail so people have  
enough context...

> Andi wrote:
>> while the delay would have been fine with me as well (for the  
>> record),
>> we didn't receive any updates on this from you, and i don't seem to  
>> be
>> able to find a review bundle either.  therefore i'd assume this will
>> (have to) be deferred until 3.2, right?
>
> I am very sorry for the delay that was caused on my part, and let me
> describe the situation a bit further, telling a bit more about the
> content of the new kss version we are about to roll out.

no problem about the delay really — we're badly lagging with the  
reviews ourselves and it's open source after all.  it'll just mean  
that it probably cannot make it into 3.1 anymore.  however, that's  
just my impression and i'd very much like to get opinions from the  
other team members and wichert about this as well.

> First of all, it contains some important stuff developed and finished
> already before last August. This includes the usage of base2 which  
> causes
> a 4-5 times clean speedup of page load according to our benchmarks
> because of faster css selection. My intention was to get these in to  
> the
> first 3.0 bugfix release, but I got into minority with my opinion  
> against
> Godefroid at that time and as a consequence those improvements are
> parking on kss.core trunk (and Plone trunk) since then.
>
> The other part of the improvements we also decided and blueprinted in
> August with the kss team. It was my task to implement these and I
> completely finished this task by the first week of January. However it
> could not be merged to trunk until Godefroid could review it, and in
> addition Godefroid made a major refactoring of the codebase which was
> only merged to trunk during January. As a consequence, I had no other
> choice than to spend almost the whole snowsprint on merging my work on
> top of Godefroid's changes, practically redoing everything manually.
> Hence the delay.

as i said, no problem here.  i think we all know what it's like...   
thanks for elaborating, btw. :)

> The reasons why I still suggest Plone 3.1 to ship with the new kss
> version are the following:
>
> - Big part of the code we ship is not last minute but quite old (in  
> fact,
> at the moment we even retargeted so many features to the _next_ kss
> version that we will be able to prerelease it during March, which  
> means
> we hopefully won't get into the same situation again.)

i don't think it matters that much how old and therefore hopefully  
mature code is in this case.  of course it usually does, but the point  
is that we need a stable bundle in time to be able to review and merge  
things without affecting the release schedule.  i know this sounds  
pretty funny considering we've only just started to get going with the  
reviews ourselves, but imho — and please correct me if i'm wrong —  
your plip will be quite a chunk to review and there's simply not  
enough time left.

> - The new code brings some new features, is somewhat faster, but most
> importantly, it is more clean and much better tested then the previous
> version. So it is actually more stable then the one currently  
> included.
>
> - It has little or no impact to the application part of Plone, other  
> then
> speedups and fixing some issues. Which means it's pretty easy to  
> swap it
> around even in the last moment, in case that becomes necessary.

still, it needs to be thoroughly reviewed and testet, which takes time  
and resources we don't seem to have.  in my view one of the  
consequences of this might have to be that we'll have to lower the  
load a little bit by being strict about the submission deadline.   
however, i also think this doesn't have to affect your release at all  
— the new version can still act as a drop-in replacement known to work  
with 3.1, it just wouldn't come with 3.1 ootb.  this way it'd be  
likely to be really well tested when it comes to 3.2.

again, that's just my opinion here — the other team members might  
think different... :)

> I will give a preview buildout today together with the changelog.  
> However
> I understand that this may not give enough time for the Framework  
> Team to
> review, and that missing the deadline in itself qualifies for a
> rejection. I also did not want a special procedure for kss, and I  
> really
> wanted to comply to the deadlines very hard. As mentioned above, the
> reasons to fail this were mostly organizational. So I would leave it  
> to
> the best consideration of the Framework Team to decide about the
> inclusion. In case of any decision made, the new kss version will be
> released together with Plone 3.1.

that's great to know, at least.  and like i said (or maybe not yet),  
those reasons are perfectly understandable.  i wanted to start  
reviewing things by the beginning of last week myself, but couldn't  
because of a couple of unforeseen things.  so no complaints  or  
anything like it here.  on the contrary, personally i'd like to see  
the new version in 3.1 — it's just that we might have to defer it for  
pragmatical reasons...

> Best wishes,

cheers,


andi

--
zeidler it consulting - http://zitc.de/ - info at zitc.de
friedelstraße 31 - 12047 berlin - telefon +49 30 25563779
pgp key at http://zitc.de/pgp - http://wwwkeys.de.pgp.net/
plone 3.0.5 released! -- http://plone.org/products/plone

-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 186 bytes
Desc: This is a digitally signed message part
URL: <http://lists.plone.org/pipermail/plone-framework-team/attachments/20080130/aba198d2/attachment.sig>