Start of build config for custom user32 import libs.

Start of build config for custom user32 import libs.

BUG=None

[cpu_(ooo_6.6-7.5), 2013-02-05 19:28:44]

So to be clear the python script does not produce user32.imports.json ?

https://codereview.chromium.org/12210017/diff/1/chrome/imports/filter_export_...
File chrome/imports/filter_export_list.py (right):

https://codereview.chromium.org/12210017/diff/1/chrome/imports/filter_export_...
chrome/imports/filter_export_list.py:42: # Slurp the master file.
I agree, eval should be renamed to slurp :) Smartass note; the json files in
this CL are not json, just python dicts but I don't know of a better extension.

https://codereview.chromium.org/12210017/diff/1/chrome/imports/filter_export_...
chrome/imports/filter_export_list.py:84: print export
how does the output look like?

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports...
File chrome/imports/user32.imports.json (right):

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports...
chrome/imports/user32.imports.json:15: "delay_imports": [
is the delay loading going to be custom? I mean done by our code?

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports...
chrome/imports/user32.imports.json:28: "disabled_imports": [
I like this part a lot

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports...
chrome/imports/user32.imports.json:120: "imports": [
I think delayed and imports dicts should be ahead of disabled.

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.win8.ex...
File chrome/imports/user32.win8.exports.json (right):

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.win8.ex...
chrome/imports/user32.win8.exports.json:8: [
so I guess this file needs to be manually created every time we update the sdk.

[Sigurður Ásgeirsson, 2013-02-05 19:39:58]

On Tue, Feb 5, 2013 at 2:28 PM, <cpu@chromium.org> wrote:

> So to be clear the python script does not produce user32.imports.json ?
>
> Nopes, as-is it's pretty labor-intensive to initially build that file, but
then it should be easy to maintain.


>
>
> https://codereview.chromium.**org/12210017/diff/1/chrome/**
>
imports/filter_export_list.py<https://codereview.chromium.org/12210017/diff/1/chrome/imports/filter_export_list.py>
> File chrome/imports/filter_export_**list.py (right):
>
> https://codereview.chromium.**org/12210017/diff/1/chrome/**
>
imports/filter_export_list.py#**newcode42<https://codereview.chromium.org/12210017/diff/1/chrome/imports/filter_export_list.py#newcode42>
> chrome/imports/filter_export_**list.py:42: # Slurp the master file.
> I agree, eval should be renamed to slurp :) Smartass note; the json
> files in this CL are not json, just python dicts but I don't know of a
> better extension.
>
> https://codereview.chromium.**org/12210017/diff/1/chrome/**
>
imports/filter_export_list.py#**newcode84<https://codereview.chromium.org/12210017/diff/1/chrome/imports/filter_export_list.py#newcode84>
> chrome/imports/filter_export_**list.py:84: print export
> how does the output look like?
>
> https://codereview.chromium.**org/12210017/diff/1/chrome/**
>
imports/user32.imports.json<https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports.json>
> File chrome/imports/user32.imports.**json (right):
>
> https://codereview.chromium.**org/12210017/diff/1/chrome/**
>
imports/user32.imports.json#**newcode15<https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports.json#newcode15>
> chrome/imports/user32.imports.**json:15: "delay_imports": [
> is the delay loading going to be custom? I mean done by our code?
>
> https://codereview.chromium.**org/12210017/diff/1/chrome/**
>
imports/user32.imports.json#**newcode28<https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports.json#newcode28>
> chrome/imports/user32.imports.**json:28: "disabled_imports": [
> I like this part a lot
>
> https://codereview.chromium.**org/12210017/diff/1/chrome/**
>
imports/user32.imports.json#**newcode120<https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports.json#newcode120>
> chrome/imports/user32.imports.**json:120: "imports": [
> I think delayed and imports dicts should be ahead of disabled.
>
> https://codereview.chromium.**org/12210017/diff/1/chrome/**
>
imports/user32.win8.exports.**json<https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.win8.exports.json>
> File chrome/imports/user32.win8.**exports.json (right):
>
> https://codereview.chromium.**org/12210017/diff/1/chrome/**
>
imports/user32.win8.exports.**json#newcode8<https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.win8.exports.json#newcode8>
> chrome/imports/user32.win8.**exports.json:8: [
> so I guess this file needs to be manually created every time we update
> the sdk.
>
>
https://codereview.chromium.**org/12210017/<https://codereview.chromium.org/1...
>

[Sigurður Ásgeirsson, 2013-02-05 19:48:51]

https://codereview.chromium.org/12210017/diff/1/chrome/imports/filter_export_...
File chrome/imports/filter_export_list.py (right):

https://codereview.chromium.org/12210017/diff/1/chrome/imports/filter_export_...
chrome/imports/filter_export_list.py:42: # Slurp the master file.
On 2013/02/05 19:28:44, cpu wrote:
> I agree, eval should be renamed to slurp :) Smartass note; the json files in
> this CL are not json, just python dicts but I don't know of a better
extension.

Yeah .py is disingenious, maybe .data?

https://codereview.chromium.org/12210017/diff/1/chrome/imports/filter_export_...
chrome/imports/filter_export_list.py:84: print export
On 2013/02/05 19:28:44, cpu wrote:
> how does the output look like?

The output from this script is just a run of sorted imports, e.g. 

AddClipboardFormatListener@4
CalculatePopupWindowPosition@20
CancelShutdown@0
ChangeWindowMessageFilter@8
ChangeWindowMessageFilterEx@16
CloseGestureInfoHandle@4
CloseTouchInputHandle@4
CreateDesktopExA@32

and so on.

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports...
File chrome/imports/user32.imports.json (right):

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports...
chrome/imports/user32.imports.json:15: "delay_imports": [
On 2013/02/05 19:28:44, cpu wrote:
> is the delay loading going to be custom? I mean done by our code?

Half and half. My plan is to generate an import library for "user32-delay.dll",
and then snip the "-delay" part and bind to user32.dll in a hook.
So, the binary will link to "user32_winxp.lib" and "user32_delay.lib", then
specify /delayload:user32-delay.dll.

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports...
chrome/imports/user32.imports.json:28: "disabled_imports": [
On 2013/02/05 19:28:44, cpu wrote:
> I like this part a lot

Yeah, it should save some time, though it's technically redundant - might be
better stashed in a single file in src/build. If so i'll move it somewhere else
after user32.dll is sorted?

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.imports...
chrome/imports/user32.imports.json:120: "imports": [
On 2013/02/05 19:28:44, cpu wrote:
> I think delayed and imports dicts should be ahead of disabled.

Ok.
This list probably ought to move to another "master" file, as it's going to be a
PITA to maintain this in more than one place.

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.win8.ex...
File chrome/imports/user32.win8.exports.json (right):

https://codereview.chromium.org/12210017/diff/1/chrome/imports/user32.win8.ex...
chrome/imports/user32.win8.exports.json:8: [
On 2013/02/05 19:28:44, cpu wrote:
> so I guess this file needs to be manually created every time we update the
sdk.

Yups, if we need it around in the end game.

[M-A Ruel, 2013-02-07 21:14:30]

Drive-by full of style nits to fake like I was still doing important code
reviews.

https://codereview.chromium.org/12210017/diff/6006/chrome/app/delay_load_hook.cc
File chrome/app/delay_load_hook.cc (right):

https://codereview.chromium.org/12210017/diff/6006/chrome/app/delay_load_hook...
chrome/app/delay_load_hook.cc:18: // object file (as opposed to indirectly
thruogh a library), these pointers
through

https://codereview.chromium.org/12210017/diff/6006/chrome/app/delay_load_hook...
chrome/app/delay_load_hook.cc:37: HMODULE dll =
::LoadLibraryA(dll_name.c_str());
return reinterpret_cast<FARPROC>(::LoadLibraryA(dll_name.c_str()));

Unless you wanted to increase your LoC for the month.

https://codereview.chromium.org/12210017/diff/6006/chrome/imports/build_impor...
File chrome/imports/build_import_libraries.py (right):

https://codereview.chromium.org/12210017/diff/6006/chrome/imports/build_impor...
chrome/imports/build_import_libraries.py:6: # A utility script to create minimal
import libs from an import description
Make that a docstring

https://codereview.chromium.org/12210017/diff/6006/chrome/imports/build_impor...
chrome/imports/build_import_libraries.py:22: Creates a pair of import libraries
from imports-file.
Often, I make this the docstring and then use:

optparse.OptionParser(
    usage='%prog [options] [imports-file]',
    description=sys.modules[__name__].__doc__)

So the description it taken right away from the module's docstring. Don't you
like that?

https://codereview.chromium.org/12210017/diff/6006/chrome/imports/build_impor...
chrome/imports/build_import_libraries.py:87: # For each library write an assmbly
file containing empty declarations
assembly

And I'd make the whole thing a docstring.