Add icu.isolate

Add icu.isolate

icu.isolate add either icudtl.dat or icudt.dll to |track...| depending on
icu_use_data_file_flag and OS.

icu.isolate will be included by *isolate files that currently list icudt.dll on
Windows.

This is a prerequisite for the following CLs to use icudtl.dat :

Windows : https://codereview.chromium.org/99473012
Linux : https://codereview.chromium.org/102413007
Mac : https://codereview.chromium.org/109013004

It's also a prerequisite for the following CLs to refer to icu.isolate :

https://codereview.chromium.org/124143002/

BUG=72633
TEST=Once rolled in, *swam testbots work fine.
R=mark@chromium.org, maruel@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=243376

[jungshik at Google, 2014-01-03 19:51:07]

can you review this CL in tandem with the following CL ? 

https://codereview.chromium.org/124143002/  

Thanks

[jungshik at Google, 2014-01-03 19:53:40]

On 2014/01/03 19:51:07, Jungshik Shin wrote:
> can you review this CL in tandem with the following CL ? 
> 
> https://codereview.chromium.org/124143002/  

The above CL (not this one) includes icu.isolate. What I'm not sure of is at
what level 'conditions' in icu.isolate will be positioned in 'including isolate
files'.  If 'conditions' are at the outermost level, it'll obscure 'conditions'
listed earlier in the 'includers'. 


> 
> Thanks

[Mark Mentovai, 2014-01-03 20:13:38]

LGTM on the syntax, but I defer to Marc-Antoine on the isolate-ness of it.

[Mark Mentovai, 2014-01-03 20:15:50]

Nothing should be obscured. Unlike writing

{
  'conditions': [ /* block 1 */ ],
  'conditions': [ /* block 2 */ ],
}

where block 2 winds up replacing block 1, when GYP merges things like include
files, *_settings, conditions, etc., it resolves things intelligently.
https://code.google.com/p/gyp/wiki/InputFormatReference#Merging

[jungshik at Google, 2014-01-03 20:57:08]

On 2014/01/03 20:15:50, Mark Mentovai wrote:
> Nothing should be obscured. Unlike writing
> 
> {
>   'conditions': [ /* block 1 */ ],
>   'conditions': [ /* block 2 */ ],
> }
> 
> where block 2 winds up replacing block 1, when GYP merges things like include
> files, *_settings, conditions, etc., it resolves things intelligently.
> https://code.google.com/p/gyp/wiki/InputFormatReference#Merging

Thanks a lot for the review and the pointer/explanation !!

[jungshik at Google, 2014-01-06 18:34:10]

On 2014/01/03 20:13:38, Mark Mentovai wrote:
> LGTM on the syntax, but I defer to Marc-Antoine on the isolate-ness of it.

Thanks. 
@maruel, can you take a look at this and the corresponding chrome-side change (
https://codereview.chromium.org/124143002/ )?

[jungshik at Google, 2014-01-06 23:57:21]

I'll try to fix this issue, but I'm not sure what to do about an assertion
failure in convert_old_to_new_format().

https://codereview.chromium.org/99073008/diff/80001/icu.isolate
File icu.isolate (right):

https://codereview.chromium.org/99073008/diff/80001/icu.isolate#newcode6
icu.isolate:6: ['icu_use_data_file_flag==1 and OS!="ios"', {
The above line is not liked by convert_old_to_new_format() in isolate.py. That's
for the backward compatibility and has a comment saying that it should be gone
when no old isolate format file is left. 

With a call to convert_old_to_new_format() commented out, I still have an
assertion failure ('len(condition)==2' in verify_condition in isolate.py). 

It looks like I'm not supposed to use if-then-else. Instead, a series of
if-then's have to be used. 

I wonder whether a nested conditions are allowed.

[M-A Ruel, 2014-01-07 14:07:43]

On 2014/01/06 23:57:21, Jungshik Shin wrote:
> I'll try to fix this issue, but I'm not sure what to do about an assertion
> failure in convert_old_to_new_format().
> 
> https://codereview.chromium.org/99073008/diff/80001/icu.isolate
> File icu.isolate (right):
> 
> https://codereview.chromium.org/99073008/diff/80001/icu.isolate#newcode6
> icu.isolate:6: ['icu_use_data_file_flag==1 and OS!="ios"', {
> The above line is not liked by convert_old_to_new_format() in isolate.py.
That's
> for the backward compatibility and has a comment saying that it should be gone
> when no old isolate format file is left. 
> 
> With a call to convert_old_to_new_format() commented out, I still have an
> assertion failure ('len(condition)==2' in verify_condition in isolate.py). 
> 
> It looks like I'm not supposed to use if-then-else. Instead, a series of
> if-then's have to be used. 
> 
> I wonder whether a nested conditions are allowed.

Exact, do not use if/else, neither nested conditions. The code enforces simple
logic. This is a trade-off between maintainability, repetitivity and just dead
simple parser.

lgtm

[jungshik at Google, 2014-01-07 19:27:49]

On 2014/01/07 14:07:43, M-A Ruel wrote:
> On 2014/01/06 23:57:21, Jungshik Shin wrote:
> > I'll try to fix this issue, but I'm not sure what to do about an assertion
> > failure in convert_old_to_new_format().
> > 
> > https://codereview.chromium.org/99073008/diff/80001/icu.isolate
> > File icu.isolate (right):
> > 
> > https://codereview.chromium.org/99073008/diff/80001/icu.isolate#newcode6
> > icu.isolate:6: ['icu_use_data_file_flag==1 and OS!="ios"', {
> > The above line is not liked by convert_old_to_new_format() in isolate.py.
> That's
> > for the backward compatibility and has a comment saying that it should be
gone
> > when no old isolate format file is left. 
> > 
> > With a call to convert_old_to_new_format() commented out, I still have an
> > assertion failure ('len(condition)==2' in verify_condition in isolate.py). 
> > 
> > It looks like I'm not supposed to use if-then-else. Instead, a series of
> > if-then's have to be used. 
> > 
> > I wonder whether a nested conditions are allowed.
> 
> Exact, do not use if/else, neither nested conditions. The code enforces simple
> logic. This is a trade-off between maintainability, repetitivity and just dead
> simple parser.
> 
> lgtm

Thank you for the review. BTW, '!=' is not allowed either, is it? I turned 'OS
!= "ios"' to 'OS == "linux" or OS == "mac" or OS == "win" or OS == "android"'.
Am I right on this?

[M-A Ruel, 2014-01-07 19:28:31]

On 2014/01/07 19:27:49, Jungshik Shin wrote:
> On 2014/01/07 14:07:43, M-A Ruel wrote:
> > On 2014/01/06 23:57:21, Jungshik Shin wrote:
> > > I'll try to fix this issue, but I'm not sure what to do about an assertion
> > > failure in convert_old_to_new_format().
> > > 
> > > https://codereview.chromium.org/99073008/diff/80001/icu.isolate
> > > File icu.isolate (right):
> > > 
> > > https://codereview.chromium.org/99073008/diff/80001/icu.isolate#newcode6
> > > icu.isolate:6: ['icu_use_data_file_flag==1 and OS!="ios"', {
> > > The above line is not liked by convert_old_to_new_format() in isolate.py.
> > That's
> > > for the backward compatibility and has a comment saying that it should be
> gone
> > > when no old isolate format file is left. 
> > > 
> > > With a call to convert_old_to_new_format() commented out, I still have an
> > > assertion failure ('len(condition)==2' in verify_condition in isolate.py).

> > > 
> > > It looks like I'm not supposed to use if-then-else. Instead, a series of
> > > if-then's have to be used. 
> > > 
> > > I wonder whether a nested conditions are allowed.
> > 
> > Exact, do not use if/else, neither nested conditions. The code enforces
simple
> > logic. This is a trade-off between maintainability, repetitivity and just
dead
> > simple parser.
> > 
> > lgtm
> 
> Thank you for the review. BTW, '!=' is not allowed either, is it? I turned 'OS
> != "ios"' to 'OS == "linux" or OS == "mac" or OS == "win" or OS == "android"'.
> Am I right on this?

Exact, only inclusion is accepted, not exclusion.

[jungshik at Google, 2014-01-07 20:41:13]

On 2014/01/07 19:28:31, M-A Ruel wrote:
> On 2014/01/07 19:27:49, Jungshik Shin wrote:
> > On 2014/01/07 14:07:43, M-A Ruel wrote:
> > > On 2014/01/06 23:57:21, Jungshik Shin wrote:
> > > > I'll try to fix this issue, but I'm not sure what to do about an
assertion
> > > > failure in convert_old_to_new_format().
> > > > 
> > > > https://codereview.chromium.org/99073008/diff/80001/icu.isolate
> > > > File icu.isolate (right):
> > > > 
> > > > https://codereview.chromium.org/99073008/diff/80001/icu.isolate#newcode6
> > > > icu.isolate:6: ['icu_use_data_file_flag==1 and OS!="ios"', {
> > > > The above line is not liked by convert_old_to_new_format() in
isolate.py.
> > > That's
> > > > for the backward compatibility and has a comment saying that it should
be
> > gone
> > > > when no old isolate format file is left. 
> > > > 
> > > > With a call to convert_old_to_new_format() commented out, I still have
an
> > > > assertion failure ('len(condition)==2' in verify_condition in
isolate.py).
> 
> > > > 
> > > > It looks like I'm not supposed to use if-then-else. Instead, a series of
> > > > if-then's have to be used. 
> > > > 
> > > > I wonder whether a nested conditions are allowed.
> > > 
> > > Exact, do not use if/else, neither nested conditions. The code enforces
> simple
> > > logic. This is a trade-off between maintainability, repetitivity and just
> dead
> > > simple parser.
> > > 
> > > lgtm
> > 
> > Thank you for the review. BTW, '!=' is not allowed either, is it? I turned
'OS
> > != "ios"' to 'OS == "linux" or OS == "mac" or OS == "win" or OS ==
"android"'.
> > Am I right on this?
> 
> Exact, only inclusion is accepted, not exclusion.

Thank you for confirming it. I'll commit this cl (the ICU-side change) and wait
for your fix for isolate.py before landing the chrome-side CL
(https://codereview.chromium.org/124143002/ ).

[jungshik at Google, 2014-01-07 20:43:15]

Committed patchset #7 manually as r243376 (presubmit successful).