Pulling changes from git://pyj.be/git/pyjamas.git

37 views
Skip to first unread message

Sarvi Shanmugham

unread,
Oct 31, 2013, 1:17:09 PM10/31/13
to pyjs-...@googlegroups.com
Hi
    I am in the process of pulling in changes from
git://pyj.be/git/pyjamas.git   master
which has also seen bug fixes and development since the branch off.

Just wanted to to make sure that people did not have any concerns in doing it.
Both are still BSD licensed open source code code with contributions from
not one but many and so I don't see any issues

If yall have any concerns in doing it do let me know.

Sarvi

Sarvi Shanmugham

unread,
Oct 31, 2013, 1:22:31 PM10/31/13
to pyjs-...@googlegroups.com
While at it do take a look at the code being merged and let me know if people 
see a problem with the code being merged as well.
I do see a whole bunch of conflicts for now, but was gonna work on it this week.

Sarvi

Glyph

unread,
Oct 31, 2013, 1:22:56 PM10/31/13
to pyjs-...@googlegroups.com
Sounds sensible to me, although it would be nice to drop the authors of those changes a note, thank them, and ask them to participate and contribute directly to the main repository in the future.

-glyph



Saravanan Shanmugham

unread,
Oct 31, 2013, 1:27:25 PM10/31/13
to pyjs-...@googlegroups.com
I've already tried that once, a few months ago. And his answer was no.
But I will definitely leave another note to Luke inviting him to work with us.

Sarvi





--

---
You received this message because you are subscribed to a topic in the Google Groups "Pyjs.org Users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/pyjs-users/jdjHf3j1_oU/unsubscribe.
To unsubscribe from this group and all its topics, send an email to pyjs-users+...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

Sarvi Shanmugham

unread,
Nov 2, 2013, 2:34:43 PM11/2/13
to pyjs-...@googlegroups.com
I went through dry run of the merge. Its been a while and there were a lot of changes.
Big ticket Items
    1. Kees's change from doted notation to dictionary notation is causing a lot of conflicts in one file.
    2. The pyjs_site directory is gone from pyjs.org repo and is still exists in pyj.be
    3. The Calendar example seems to have gone through development on both sides and from what I can tell for the same feature.
    4. The directory restructure on pyjs.org side though git seems to be smart about merging it well.
    5. A long list of bug fixes on both sides.

My approach for the final run.
    1. pick up the merges that are automatic.
    2. directory structure and pyjs_site removal leave as is for pyjs.org
    3. Calendar example exclude pyj.be changes since pyjs.org seems to have the same functionality an is cleaner.
    4. dot notation vs dictionary notation - I am not sure. From playing around with closure compiler options. Looks like I can compile the closure compiler
and eliminate property renaming in the advanced_optimizations, which will still meet what Kees was trying to do with his changes, keep the generated code readable 
and also avoid a lot of conflicts. So this is leading me to think I should restore it back to the doted notation and make the closure compiler changes to match it
so that Kees's goals are still met. and then try the merge with pyj.be after. would reduce the conflicts from what I can tell.

Kees, what do you think?

Sarvi
To unsubscribe from this group and all its topics, send an email to pyjs-users+unsubscribe@googlegroups.com.

Kees Bos

unread,
Nov 2, 2013, 4:27:01 PM11/2/13
to pyjs-...@googlegroups.com
Two things:

1. When you ask someone whether you may use his code and he says 'No',
then you shouldn't (regardless of the license, that's my opinion).
2. If you manage to get closure working, I don't care about dotted
notation or not. If you don't get closure working, I rather not go back
to dotted notation. So, try to get closure working first (before you go
back to dotted notation). Note that it'll have to work for dynamic
loading. If I recall well, closure will change the code to dotted
notation anyway (even without maximal optimization).
> +unsub...@googlegroups.com.
> For more options, visit
> https://groups.google.com/groups/opt_out.
>
>
>
> --
>
> ---
> You received this message because you are subscribed to the Google
> Groups "Pyjs.org Users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to pyjs-users+...@googlegroups.com.

Sarvi Shanmugham

unread,
Nov 2, 2013, 4:43:56 PM11/2/13
to pyjs-...@googlegroups.com
1. I didn't ask if I could use his code. Which I will ask explicitly again. I had just encouraged him a few months ago to set aside his differences work together for which he was not interested. Which I do not interpret as no to merging. But will check again.
2. I will get closure working before goin there. Is there an example that deals with this dynamic loading specifically that I can test with? I recently removed code that was importing dynamic which was causing pyjs tools to have dependency on gwt. It didn't break any test cases and the examples ran fine.

Sarvi

Kees Bos

unread,
Nov 2, 2013, 5:01:10 PM11/2/13
to pyjs-...@googlegroups.com
On Sat, 2013-11-02 at 13:43 -0700, Sarvi Shanmugham wrote:
> 1. I didn't ask if I could use his code. Which I will ask explicitly again. I had just encouraged him a few months ago to set aside his differences work together for which he was not interested. Which I do not interpret as no to merging. But will check again.
OK

> 2. I will get closure working before goin there. Is there an example that deals with this dynamic loading specifically that I can test with? I recently removed code that was importing dynamic which was causing pyjs tools to have dependency on gwt. It didn't break any test cases and the examples ran fine.

It's a build flag, I don't recall the exact word (try --help). It make
the builder create a smaaalll .html and the lib/*.js files will be
loaded separately.

Łukasz Mach

unread,
Nov 4, 2013, 9:19:57 AM11/4/13
to pyjs-...@googlegroups.com
W dniu 02.11.2013 19:34, Sarvi Shanmugham pisze:
> I went through dry run of the merge. Its been a while and there were a
> lot of changes.
> Big ticket Items
> 1. Kees's change from doted notation to dictionary notation is
> causing a lot of conflicts in one file.
> 2. The pyjs_site directory is gone from pyjs.org repo and is still
> exists in pyj.be
> 3. The Calendar example seems to have gone through development on
> both sides and from what I can tell for the same feature.

I'm a bit involved into Calendar/Datefield/Montfield. So words "Calendar
example" triggered me that it's maybe about something mine. However - I
don't understand what do you mean. What does mean that it have gone
through development on both sides?

--
pozdrawiam

Łukasz Mach - lukas...@pagema.net

Sarvi Shanmugham

unread,
Nov 4, 2013, 2:11:25 PM11/4/13
to pyjs-...@googlegroups.com
During the trial merge I found saw both conflicts as well non merge-conflicting code that
that seemed like was addressing the same Calendar stuff.

You can see those changes here https://github.com/sarvi/pyjs/commits/www.pyj.be
Look in the June timeframe of the commits.

This is branch is tracking www.pyj.be on github and I thought I will keep it around as a place to 
continue this merging/tracking work as long as it is feasible. My guess it will these will probably diverge 
to point where it doesnt make sense anymore. But until then its worth exchanging bug fixes and I hope this is vehicle
for it.

during my dry run, it must my understanding that both sides were tackling similar problems
and  I kept the pyjs.org version(which I presume is your changes) instead of the pyj.be versions.
Thought yours were cleaner. I could be wrong and they could be completely different features.

If you can merge the code to the point where the Calendar stuff was commited, that would be helpfull.
Or, you can help review my merge when I start it. I am not planning on working on the merge
until some of kees's concerns with the closure compiler advanced are addressed.

Sarvi  

Sarvi Shanmugham

unread,
Nov 4, 2013, 2:23:26 PM11/4/13
to pyjs-...@googlegroups.com
Reply all
Reply to author
Forward
0 new messages