Re: Feedback Needed! Changing how Cheetah handles imports

On Thu, Apr 16, 2009 at 04:13:00PM -0700, [email protected] wrote:
>    My personal vote is for #2 being the default behavior since it follows
>    Python importing semantics closely (we also have updated our templates
>    at Slide to avoid the issue Martin Skott describes).=20
>     
>    How do u do this? ie ensure all your 300 templates dont have this issue
>    cuz we have tons of templates but we havent figured out how to test all
>    the templates for such corner cases


We employ some stellar QA engineers in addition to a large suite of
PyUnit tests. We basically scheduled a period of a week prior to rolling
the new version of Cheetah to test it on our development and continuous
integration machines (p.s. we're hiring :-P)


Cheers



>    On Thu, Apr 16, 2009 at 2:34 PM, R. Tyler Ballance <[email protected]>
>    wrote:
> 
>      Since this issue has come up (see thread started by Martin Skott) with
>      regards to how newer versions of Cheetah are handling the #import
>      statement.
> 
>      In CVS versions of Cheetah (i.e. v2.0.1 and earlier), all #import
>      directives were collected and *moved* to the top of the generated Python
>      code for the template. This means things like this:
> 
>             #import sys
>             ## This is a crazy template
>             #set $foo = 'bar'
>             #import cjson
> 
>             #def method()
>                     #import simplejson
>                     #return simplejson.dumps({'foo' : 'bar'})
>             #end def
> 
>      Generates code like this:
> 
>             import sys
>             import cjson
>             import simplejson
> 
>             class test(Template):
>                     def writeBody(self):
>                             ## This is a crazy template
>                             foo = 'bar'
> 
>                     def method(self):
>                             return simplejson.dumps({'foo' : 'bar'})
> 
>      In Cheetah v2.1.xx this behavior has been changed to support proper
>      inline imports (I think the "magic global import namespace" is a bad
>      idea and restricts the power of what you can do within Cheetah)
> 
>      The code above would generate something like this instead:
> 
>             import sys
> 
>             class test(Template):
>                     def writeBody(self):
>                             ## This is a crazy template
>                             foo = 'bar'
>                             import cjson
> 
>                     def method(self):
>                             import simplejson
>                             return simplejson.dumps({'foo' : 'bar'})
> 
>      There are a couple things at play here, firstly anything outside of a
>      defined function, effectively falls into the "default method" which is
>      writeBody(). Secondly I think it's important to support inline imports
>      the same way Cheetah supports generate scoping of variables.
> 
>      I think the only solution we have is to use a compile-time setting for
>      this, it's quite the "damned if you do, damned if you don't" scenario.
> 
>      The decision to be made between "us" as the community is which we want
>      to be the *default* behavior:
>             1: Non-scoping import behavior (2.0.1 and earlier)
>             2: Scoping/inline import behavior.(2.1.0 and later)
> 
>      My personal vote is for #2 being the default behavior since it follows
>      Python importing semantics closely (we also have updated our templates
>      at Slide to avoid the issue Martin Skott describes).
> 
>      What say ye all? :)
>      --
>      -R. Tyler Ballance
>      Slide, Inc.
>      ------------------------------------------------------------------------------
>      Stay on top of everything new and different, both inside and
>      around Java (TM) technology - register by April 22, and save
>      $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
>      300 plus technical and hands-on sessions. Register today.
>      Use priority code J9JMT32. http://p.sf.net/sfu/p
>      _______________________________________________
>      Cheetahtemplate-discuss mailing list
>      [email protected]
>      https://lists.sourceforge.net/lists/listinfo/cheetahtemplate-discuss
> 
>    --
>    Bidegg worlds best auction site
>    http://bidegg.com

-- 
-R. Tyler Ballance
Slide, Inc.

------------------------------------------------------------------------------
Stay on top of everything new and different, both inside and 
around Java (TM) technology - register by April 22, and save
$200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
300 plus technical and hands-on sessions. Register today. 
Use priority code J9JMT32. http://p.sf.net/sfu/p

_______________________________________________
Cheetahtemplate-discuss mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/cheetahtemplate-discuss

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.9 (GNU/Linux)

iEYEARECAAYFAknnvn8ACgkQFCbH3D9R4W/7yQCfbQbeiAhzBMga6QHc/p4gGfto
MW8AnRVRZZusKUOhyHqFpYtjYpxq+y8R
=lbXx
-----END PGP SIGNATURE-----