Python UK meeting and Django

[Kevin M. Smith]

Quoting Kapil Thangavelu <kapil.foss@...>:

> On Thu, May 8, 2008 at 3:54 PM, Kevin Smith <kevin@...> wrote:
>
>>
>> Noah Gift wrote:
>>
>>
>>> They were successful due to brilliant marketing (kudos man), a
>>> mostly-developed python answer to RoR, a good sense of timing and strong
>>> community development skills.
>>
>>
>>  I agree, and if you doubt this truth, then read Four Hour Work Week.
>>  Anyone can create their own image, and expertise in weeks, but it requires
>> tenacious and persistent marketing.  They, or the entity, needs to write
>> many articles, give talks, etc.  Unfortunately, often the very smartest
>> people suck the most at engaging in this type of necessary behavior.
>>
>>
>>>
>>>
>>> Perhaps grok is being too nice. GROK SMASH YET ANOTHER ROR RIPOFF.
>>> </rant>
>>>
>>> Of course a real advantage for DJango is that they are AppEngine ready.
>>> Interestingly the Datastore is much more akin to the ZODB than it is to an
>>> RDB.
>>
>>
>>  I have been playing with the Datastore and like it quite a bit.  I would
>> love to see a version of Grok that just uses the Datastore as option.  I
>> would really, really like Grok then.  I know we have a "Global Sprint"
>> coming at the end of May, and also some people that use Grok will be
>> attending the Google I/O event.  Anyone want to try...?
>>
>>
>> Yup, I'm game. The 1,000 file limit is pretty tough. A did a basic test
>> doing an install of ore.wsgiref, and ended up with 3000+ files without
>> including zcml.
>>
>> My mini-strategy is to
>> * get easy_install zope.component
>> * get martian working
>> * implement custom traversal
>> * fork grok  :) doh!
>>
>> After looking at the Datastore it may yet  be possible to do object-based
>> publishing by replacing persistence. I'll have to catch up with you at
>> Google I/O.
>>
>> fwiw, i started playing around with trying get z3 onto gae, its a hard
> issue, after fiddling away the proxies in zope.deferredmodule, there's still
> pkg_resource fixes to get just a barebones zope.publisher.paste running, and
> then there are issues with location.proxies.. you'll end up with a very
> stripped down z3, with little resemblance to the standard stack just pure
> object publishing, and thats not even including a templating engine, which
> have proxies referenced in them. and of course the 1000 file limit is tough,
> to work around you can try zip'd files but that kills existing zcml usage
> and template usage. at that point i realized i was just reinventing what the
> repoze guys have been doing.. and that for running a z3 like environment the
> repoze stack (obob publisher, transaction) is likely to be quite a bit
> easier to just run.
>
> cheers,
>
> kapil
>

Nice work. Off-hand do you think it would it be more feasible to port  
Zope2/Repoze/five.grok?



_______________________________________________
Grok-dev mailing list
Grok-dev@...
http://mail.zope.org/mailman/listinfo/grok-dev

[TIm Terlegård-3]

On May 8, 2008, at 10:07 PM, Noah Gift wrote:

>> I have been playing with the Datastore and like it quite a bit.  I  
>> would love to see a version of Grok that just uses the Datastore as  
>> option.  I would really, really like Grok then.  I know we have a  
>> "Global Sprint" coming at the end of May, and also some people that  
>> use Grok will be attending the Google I/O event.  Anyone want to  
>> try...?
>
> Yup, I'm game. The 1,000 file limit is pretty tough. A did a basic  
> test doing an install of ore.wsgiref, and ended up with 3000+ files  
> without including zcml.
>
> My mini-strategy is to
> * get easy_install zope.component
> * get martian working
> * implement custom traversal
> * fork grok  :) doh!
>
> After looking at the Datastore it may yet  be possible to do object-
> based publishing by replacing persistence. I'll have to catch up  
> with you at Google I/O.

Isn't there a restriction about long running processes? Zope needs
to run as a service or it will have to reregister all adapters/utilities
on each request. How is this solved?

/Tim
_______________________________________________
Grok-dev mailing list
Grok-dev@...
http://mail.zope.org/mailman/listinfo/grok-dev

[Noah Gift]

On Fri, May 9, 2008 at 1:49 AM, Tim Terlegård <tim.terlegard@...> wrote:

On May 8, 2008, at 10:07 PM, Noah Gift wrote:

I have been playing with the Datastore and like it quite a bit.  I would love to see a version of Grok that just uses the Datastore as option.  I would really, really like Grok then.  I know we have a "Global Sprint" coming at the end of May, and also some people that use Grok will be attending the Google I/O event.  Anyone want to try...?

Yup, I'm game. The 1,000 file limit is pretty tough. A did a basic test doing an install of ore.wsgiref, and ended up with 3000+ files without including zcml.

My mini-strategy is to
* get easy_install zope.component
* get martian working
* implement custom traversal
* fork grok  :) doh!

After looking at the Datastore it may yet  be possible to do object-based publishing by replacing persistence. I'll have to catch up with you at Google I/O.

Isn't there a restriction about long running processes? Zope needs
to run as a service or it will have to reregister all adapters/utilities
on each request. How is this solved?

Can Zope run under CGI?  I like the idea of starting at the problem backwards and getting anything at all of any sort to run in whatever hackish way possible, and then slowly normalizing it. 

/Tim

_______________________________________________
Grok-dev mailing list
Grok-dev@...
http://mail.zope.org/mailman/listinfo/grok-dev

[Kapil Thangavelu-5]

On Fri, May 9, 2008 at 1:49 AM, Tim Terlegård <tim.terlegard@...> wrote:

On May 8, 2008, at 10:07 PM, Noah Gift wrote:

I have been playing with the Datastore and like it quite a bit.  I would love to see a version of Grok that just uses the Datastore as option.  I would really, really like Grok then.  I know we have a "Global Sprint" coming at the end of May, and also some people that use Grok will be attending the Google I/O event.  Anyone want to try...?

Yup, I'm game. The 1,000 file limit is pretty tough. A did a basic test doing an install of ore.wsgiref, and ended up with 3000+ files without including zcml.

My mini-strategy is to
* get easy_install zope.component
* get martian working
* implement custom traversal
* fork grok  :) doh!

After looking at the Datastore it may yet  be possible to do object-based publishing by replacing persistence. I'll have to catch up with you at Google I/O.

Isn't there a restriction about long running processes? Zope needs
to run as a service or it will have to reregister all adapters/utilities
on each request. How is this solved?

via import caching, and finishing component loading/registration at a module global scope.

http://code.google.com/appengine/docs/python/appcaching.html

cheers,

kapil

_______________________________________________
Grok-dev mailing list
Grok-dev@...
http://mail.zope.org/mailman/listinfo/grok-dev

[Lennart Regebro-2]

On Fri, May 9, 2008 at 5:07 AM,  <kevin@...> wrote:
> Nice work. Off-hand do you think it would it be more feasible to port
> Zope2/Repoze/five.grok?

The bet strategy to get Zope2 running on google app engine is to first
make Zope2 a Zope3 application, and then port Zope 3. :)

--
Lennart Regebro: Zope and Plone consulting.
http://www.colliberty.com/
[phone]
_______________________________________________
Grok-dev mailing list
Grok-dev@...
http://mail.zope.org/mailman/listinfo/grok-dev