[Laszlo-dev] Finding the change that's causing problems

[Max Carlson]

(redirecting to laszlo-dev so other folks can learn about this process - 
please post to the forum as well if you'd like!)

Hi Robert,

Sorry to hear that!  We'll need to get this isolated to a testcase we 
can reproduce locally to fix.  I can't think of any changes off hand 
that would have caused this.  There are two options:

a) Manually disable/comment out parts of your application until you can 
isolate the part that shows the problem.  This is best if your 
application is relatively small, or you don't have a build environment 
set up.

b) Use the binary search tool 
(http://svn.openlaszlo.org/tools/trunk/svn/binsearch.rb) to isolate the 
revision that is causing the problem.  This is probably a better option 
since you know things worked in 4.2, but broke since then - but you need 
to be able to build from scratch.

4.2/4.2.0.1 is a subset of trunk, so this should go fairly quickly. 
First, you'll need to pull down a copy of the 4.2 branch, at 
http://svn.openlaszlo.org/openlaszlo/branches/4.2/.  I recommend syncing 
to r12212 which is when 4.2.0 was tagged according to 
http://svn.openlaszlo.org/openlaszlo/tags/4.2.0/.  From there, run 
through the revisions up to r12833 or 
http://svn.openlaszlo.org/openlaszlo/tags/4.2.0.1/.

Here's how I would do it locally if I had your app:
$ binsearch.rb -r 12211 -l 12900 .
Searching between 12305 and 12833.
Synchronizing and building 12305.  33 remaining.

Don't worry, it's not many revisions (33).  It does a binary search, so 
you can expect it to find the version that's causing you problems within 
~5 builds.

This is the same process I use when I'm not sure what build caused a 
problem.  It's always been more productive than guessing at revisions - 
  I've been surprised many times because the change that caused the 
problem almost always wasn't what I thought! 
http://www.openlaszlo.org/jira/browse/LPP-7295 has an example of me 
using binsearch.rb to isolate a bug.

Once you've done this, we'll know exactly which revision is the culprit 
and we can go from there.  Let me know if you have questions!

> Max,
> 
> I tried to test one of my apps on trunk 13070 swf8 to see if the JavaRPC XML is fixed or not.
> 
> With 4.2 swf8, the app would at least boot and receive data from the server. It looked and felt similar to the app from 4.0.12, minus data flowing.
> 
> With trunk swf8, it is much worse. The UI doesn't render and the debugger chokes out tons of "reference to undefined variable" and "undefined object does not have a property" errors immediately. Some investigating of the classes mentioned by the errors yields that although the class names appear in the "size profile" info from the dev console, classes are missing in the "lz" namespace. Some of my classes appear, but many do not.
> 
> Then, I decided to try 4.2.0.1, only to find it is broken just like trunk. Whatever changed has happened after 4.2.
> 
> So, whether or not JavaRPC XML works in trunk is academic for now, b/c my apps are totally dead on 4.2.0.1 and after.
> 
> Know of any changes in 4.2.0.1 that would not bake all my classes into the "lz" namespace? http://www.openlaszlo.org/jira/browse/LPP-7589 in 4.2.0.1 is the only one that looks suspicious to me.
> 
> Err 

-- 
Regards,
Max Carlson
OpenLaszlo.org