[reportlab-users] patch to make Reportlab work again on Jython

Raphaël Valyi rvalyi at gmail.com
Thu Feb 5 09:01:47 EST 2009


Hi Robin, inlined responses,

On Thu, Feb 5, 2009 at 12:10 PM, Robin Becker <robin at reportlab.com> wrote:

> Raphaël,

>

> we recently introduced some ability for selected people to access and modify

> at least parts of the reportlab repository. Would it not make sense for us

> to create a jython branch based on the current reportlab 2.3 and then allow

> you (and perhaps others) to make patches to that branch?


Ok, will do that at least for the current patch I had.


>

> It seems wrong to mess with the base implementation just because the latest

> jython has some limits. I suppose jython's new model(ie not creating java

> files) has caused these problems with class size, but I'm sure you'll find

> other areas where our code needs some tweaks to make it work under jython.


Well may be. I mean if you imagine to release a Jython version at
least along with the Python one that would make sense. Moreover, I
think you would have to bother propagating your Reportlab improvements
on the Jython branch because I unfortunately I'm afraid I'll never
assume that. I'm only helping out in my spare time in making the
things I believe are right, but you can only expect me to commit that
pacth and may be simple others as needs occur.
I the mean time if you can't assume to maintain that parallel Jython
distro, I think it's just easier to include the minor tweak in the
regular distro to make it work.
There might be more business around Jython in the near future
(depending a lot on what the Sun team can achieve), but currently I'm
not sure it will generate enough to sustain large efforts.



> Even if you don't want to do the commits yourself it seems to make more

> sense than doing everything on the trunk.


Well you have my view. On the other way, I can make the commit myself
and even see if the tweak could be smaller. In fact I saw that a
method was too big and didn't bother trying/guessing what was the
minimal change, instead I just wrapped every table/list instantiation
here to make it work. May be just one or two wrappings could do the
job. Indeed, if that worked before (as you seem to tell), knowing that
Jython always had this JVM limitation, I guess this is because that
particular file was simply smaller (had no time to confirm though).
That would mean that only a few wrapping would make it work. In such a
case, may be that tweak would be so minor that it could perfectly be
in the regular distro.

Meanwhile, if you Reportlab want to bring the Java Reportlab
accelerator back to life, I think it definitely makes sense to have a
separate branch + distro. Ideally thinks would be modular and clean so
that the difference between the two branches would be minimal and
auto-magically propagated.
In the ideal world that might be much like the JRuby on Rails world.
Rails started ignoring JRuby (while Sun started ignoring Rails). Now
Sun is fare from ignoring Rails (their Kenai dev platform is just
meant to be the regular corporate dev platform while it was a personal
proto of some JRuby employee, now Glassfish, the fame J2EE server is
just to be developed on... ...Kenai). And now Rails itself started
distributing the database JDBC adapter in the regular Rails distro...

As I said the same will probably happen with Jython (I see no reason
why not), but it might simply take more time.


As I said, my own goal is to show the world that a killer ERP
(OpenERP) which is using Reportab can be and should be worth running
in a Java servlet containter (Modjy). I won't bother making it happen
alone in my spare time past this stage. But I'm very much implied
professionally with OpenERP integration with increasing demand, so
it's just about bridging the gap sooner or later, if that includes
giving a hand to reportlab, I would be happy to do my part.

Best regards,

Raphaël Valyi.


>

>

> Raphaël Valyi wrote:

>>

>> Dear Reportlab developers,

>

> ........

>>

>> What is your position regarding to this?

>>

> --

> Robin Becker

> _______________________________________________

> reportlab-users mailing list

> reportlab-users at reportlab.com

> http://two.pairlist.net/mailman/listinfo/reportlab-users

>



More information about the reportlab-users mailing list