Rename deprecated "supplemental_dependencies" and clean up IDL dependency script

Rename deprecated "supplemental_dependencies" and clean up IDL dependency script

This is a cleanup patch:
* Rename "supplemental_dependencies" to "interface_dependencies",
  since these are actually partial interfaces and implements,
  not supplementals (which has a technical meaning that's not being used here)
  (Changed throughout.)

* Rename preprocess_idls.py to compute_dependencies.py,
  since this computes dependencies, and is not a preprocessor
  (a preprocessor changes source files before they hit the compiler,
   while this doesn't change the .idls: it just computes dependencies)

* Significant refactoring to compute_dependencies.py (was: preprocess_idls.py)

When renaming variables, I had trouble understanding what the variables
actually were (what's the difference between "supplemental_dependencies"
and "supplementals"?), so I cleaned up to make it clearer.

The main function parse_idl_files was actually computing a backwards
dictionary (!), and then reversing it at the end (!), which is brain-bending,
and dates back to Perl when we just had partial dependencies, not implements.
I rewrote it to just compute the dependencies in the forward order.

I also made the code simpler (less deeply nested, less hard-coded) and clearly
separated the IDL parsing from the output file writing: previously some files
were written in the middle of parsing, while others were done afterwards.

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=153766

[Nils Barth (inactive), 2013-07-09 05:04:31]

Hi haraken and/or Christophe,
Here's a cleanup CL as part of renaming "supplemental_dependencies".

...since we don't want to use "supplemental" in new code, as per:
Issue 18190004: Add Python flow to bindings generation, move dummy-generating
IDL files over
https://codereview.chromium.org/18190004/

Not urgent (just cleanup, not blocking anything), but clarifies the code.

(Some detailed comments attached.)

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/derived_so...
File Source/bindings/derived_sources.gyp (right):

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/derived_so...
Source/bindings/derived_sources.gyp:106: 'action_name':
'compute_interface_dependencies',
(Also make naming convention consistent: remove only example of camelCase.)

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/derived_so...
Source/bindings/derived_sources.gyp:118:
'<(SHARED_INTERMEDIATE_DIR)/interface_dependencies.txt',
.txt b/c text file (indeed Space Separated Values); better extension?

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/Co...
File Source/bindings/scripts/CodeGeneratorV8.pm (right):

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/Co...
Source/bindings/scripts/CodeGeneratorV8.pm:123: my $interfaceIdlFiles;
"Dependent" is inaccurate, since this starts with the main IDL file.
Instead, it's "main + partials", so it's the files for the interface;
see generate-bindings.pl.

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
File Source/bindings/scripts/compute_dependencies.py (left):

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:124: value = match.group(2) or
'VALUE_IS_MISSING'
str.partition() does exactly what we want - no need to reinvent the wheel. (>.<)

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:192: idl_file_to_interface_name
= {}
idl_file_to_interface_name: unused, removed

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:194: supplementals = {}
supplemental_dependencies/supplementals are confusing;
supplemental_dependencies is a backwards dictionary,
so remove and just have the forward dictionary.

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:203: interface_to_file = {}
Redundant with interface_name_to_idl_file, removed.

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:264:
generate_global_constructors_partial_interface("Window",
window_constructors_filename, window_constructor_attributes_list,
only_if_changed=only_if_changed)
This mixes generating output files with parsing the IDLs; refactor.

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:283: del supplementals[idl_file]
This 'del' was originally for circular dependencies in partial interfaces, hence
the "Should never occur" comment.
When 'implements' was added, this became necessary,
as it means that an interface that is implemented by another interface
should *not* have bindings generated for itself.
I've updated the comment in the revised version.

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
File Source/bindings/scripts/compute_dependencies.py (right):

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:2: #
For consistency, add executable if runs 'main()'
(...and remove from list_idl_files_with_partial_interface.py
b/c that's only loaded as a module from GYP).

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:111: raise
IdlBadFilenameError("Identifier on the left of the 'implements' statement should
be %s in %s.idl, but found %s" % (interface_name, interface_name,
match.group(1)))
"assert" is only for internal consistency (debug statements):
it can be turned off (via -O), and shouldn't be used to check bad data.
Instead, raise exception.
http://wiki.python.org/moin/UsingAssertionsEffectively

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:174: def
generate_event_names_file(destination_filename, event_names, only_if_changed):
only_if_changed is really important and omitting it has caused problems in the
past, so make it mandatory.

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:179: for filename,
extended_attributes in sorted(event_names.iteritems()):
More sorting for consistency (below too).

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:201: def
parse_idl_files(idl_files, global_constructors_filenames):
This is the main content of this CL.
Other than sorting the dependency list, output is identical.
However, the logic is different (I've simplified it);
the earlier logic was a bit twisty.

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:357:
generate_event_names_file(options.event_names_file, event_names,
options.write_file_only_if_changed)
Write all the files in one place.

https://codereview.chromium.org/18341025/diff/1001/Source/core/scripts/list_i...
File Source/core/scripts/list_idl_files_with_partial_interface.py (left):

https://codereview.chromium.org/18341025/diff/1001/Source/core/scripts/list_i...
Source/core/scripts/list_idl_files_with_partial_interface.py:2: #
Not executable, just a module
(called by pymod_do_main in GYP, and no main()).

[haraken, 2013-07-09 05:21:38]

Looks like good refactoring. LGTM.

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/derived_so...
File Source/bindings/derived_sources.gyp (right):

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/derived_so...
Source/bindings/derived_sources.gyp:129:
'<(SHARED_INTERMEDIATE_DIR)/interface_dependencies.txt',

Nit: InterfaceDependencies.txt

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
File Source/bindings/scripts/compute_dependencies.py (right):

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:202: """Returns dependencies
between IDL files, global interfaces, and events.

Nit: global interfaces => constructors on global objects

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:213: dict of global interfaces
-> list of attributes in that interface

Ditto.

[Nils Barth (inactive), 2013-07-09 06:00:40]

Thanks for review, revised and committing!

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/derived_so...
File Source/bindings/derived_sources.gyp (right):

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/derived_so...
Source/bindings/derived_sources.gyp:129:
'<(SHARED_INTERMEDIATE_DIR)/interface_dependencies.txt',
On 2013/07/09 05:21:38, haraken wrote:
> Nit: InterfaceDependencies.txt

Got it:
CamelCase.idl  <<  data
lowercase_with_underscores.py  <<  code

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
File Source/bindings/scripts/compute_dependencies.py (right):

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:202: """Returns dependencies
between IDL files, global interfaces, and events.
On 2013/07/09 05:21:38, haraken wrote:
> Nit: global interfaces => constructors on global objects

Got it, changed here and throughout!
(I wasn't terribly clear on what these actually were.)

https://codereview.chromium.org/18341025/diff/1001/Source/bindings/scripts/co...
Source/bindings/scripts/compute_dependencies.py:266:
attribute_list.extend(constructor_list)
One more minor cleanup.

[commit-bot: I haz the power, 2013-07-09 06:00:59]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/nbarth@chromium.org/18341025/5002

[commit-bot: I haz the power, 2013-07-09 06:40:40]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/nbarth@chromium.org/18341025/19001

[commit-bot: I haz the power, 2013-07-09 07:35:53]

Retried try job too often on linux_layout_rel for step(s) webkit_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_layo...

[commit-bot: I haz the power, 2013-07-09 07:42:14]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/nbarth@chromium.org/18341025/19001

[commit-bot: I haz the power, 2013-07-09 07:58:35]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/nbarth@chromium.org/18341025/19001

[commit-bot: I haz the power, 2013-07-09 09:01:53]

Change committed as 153766

[do-not-use, 2013-07-09 10:44:55]

On 2013/07/09 09:01:53, I haz the power (commit-bot) wrote:
> Change committed as 153766

Looks like bindings tests are broken:

$ Tools/Scripts/run-bindings-tests 
python: can't open file 'bindings/scripts/preprocess_idls.py': [Errno 2] No such
file or directory

Failed to generate a supplemental dependency file.

[Nils Barth (inactive), 2013-07-09 11:54:04]

On 2013/07/09 10:44:55, Christophe Dumez wrote:
> On 2013/07/09 09:01:53, I haz the power (commit-bot) wrote:
> > Change committed as 153766
> 
> Looks like bindings tests are broken:

Oops! Sorry... m(_ _)m
Fix posted:
Issue 18425003: Fix run-bindings-tests post-cleaning
https://codereview.chromium.org/18425003/