Cleanup NativeMetafile (win) interface and EMF class.

Cleanup NativeMetafile (win) interface and EMF class.

- Rename CreateDc to Init() and remove unused argument (all non-test calls were CreateDc(NULL, NULL).  [This matches cross platform interface.]
- Remove CreateFileBackedDc from the NativeMetafile interface and make InitToFile() in the EMF class.
- Remove CreateFromFile from the NativeMetafile interface and make it InitFromFile() in the EMF class.
- Move the CloseEmf method into the destructor, making the Emf class a use once class (matches actual use).

BUG=NONE
TEST=NONE

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=78666
Reverted: http://src.chromium.org/viewvc/chrome?view=rev&revision=78669

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=78969

[Lei Zhang, 2011-03-15 07:19:08]

Is this going to conflict with http://codereview.chromium.org/6611032/ ?

[vandebo (ex-Chrome), 2011-03-15 07:56:41]

Certainly, but that one can go in first. And it'll be a minor rebase for
this cl.
On Mar 15, 2011 12:19 AM, <thestig@chromium.org> wrote:

[Lei Zhang, 2011-03-16 03:14:52]

Care to rebase?

Also, there's a try bot failure in EmfTest.DC.

[vandebo (ex-Chrome), 2011-03-16 06:35:53]

Rebase wasn't bad, but the unit test is still failing.  I changed the way the
unit test works because I removed a method it was using, so I have to figure out
why it broken.  I'll figure it out and get back to you.

[vandebo (ex-Chrome), 2011-03-17 03:35:09]

All works now, ready for review.

[M-A Ruel, 2011-03-17 13:19:21]

I think your 'e' is malfunctional, see description (two times)

http://codereview.chromium.org/6695013/diff/8023/chrome/service/service_utili...
File chrome/service/service_utility_process_host.h (right):

http://codereview.chromium.org/6695013/diff/8023/chrome/service/service_utili...
chrome/service/service_utility_process_host.h:60: const printing::Emf& metafile,
I am kind of surprised that you switched back from NativeMetafile to Emf, what's
the exact purpose?

http://codereview.chromium.org/6695013/diff/8023/printing/emf_win.h
File printing/emf_win.h (right):

http://codereview.chromium.org/6695013/diff/8023/printing/emf_win.h#newcode41
printing/emf_win.h:41: virtual bool Init();
By style code, we don't allow function overloads. Use a different name on one of
the methods.

[vandebo (ex-Chrome), 2011-03-17 23:38:06]

> I think your 'e' is malfunctional, see description (two times)

Done.

http://codereview.chromium.org/6695013/diff/8023/chrome/service/service_utili...
File chrome/service/service_utility_process_host.h (right):

http://codereview.chromium.org/6695013/diff/8023/chrome/service/service_utili...
chrome/service/service_utility_process_host.h:60: const printing::Emf& metafile,
On 2011/03/17 13:19:21, Marc-Antoine Ruel wrote:
> I am kind of surprised that you switched back from NativeMetafile to Emf,
what's
> the exact purpose? 

This method doesn't really want to render to a metafile, it wants to render to
an EMF (we already have PDF data and we want to be able to print it).  This work
is in preparation for an additional metafile implementation that internally uses
PDFs. Furthermore, the methods of the metafile that this method wants to use are
becoming EMF specific (they don't apply to a PDF implementation of metafile).

http://codereview.chromium.org/6695013/diff/8023/printing/emf_win.h
File printing/emf_win.h (right):

http://codereview.chromium.org/6695013/diff/8023/printing/emf_win.h#newcode41
printing/emf_win.h:41: virtual bool Init();
On 2011/03/17 13:19:21, Marc-Antoine Ruel wrote:
> By style code, we don't allow function overloads. Use a different name on one
of
> the methods.

I was copying the bad existing pattern, Fixed.

And during the fix, I realized that CreateFileBackedDc is actually used. Also
Fixed.

[M-A Ruel, 2011-03-18 00:09:07]

lgtm

[vandebo (ex-Chrome), 2011-03-18 17:19:57]

Marc-Antoine, The Vista Tests bot did not like this change.  The printing unit
tests timed out and (presumably) made the bot unstable: 
http://build.chromium.org/p/chromium/builders/Vista%20Tests%20%281%29/builds/...
Do you have any suggestions for tracking down the problem?

[M-A Ruel, 2011-03-18 17:33:17]

1. build the test in release and run the test on a vista box/vm.
2. use the win_rel try builder with -T printing_unittests. Make sure to use
-c/--clobber because they probably need it these days.

M-A

Le 18 mars 2011 13:19, <vandebo@chromium.org> a écrit :

> Marc-Antoine, The Vista Tests bot did not like this change.  The printing
> unit
> tests timed out and (presumably) made the bot unstable:
>
>
http://build.chromium.org/p/chromium/builders/Vista%20Tests%20%281%29/builds/...
> Do you have any suggestions for tracking down the problem?
>
>
>
> http://codereview.chromium.org/6695013/
>

[vandebo (ex-Chrome), 2011-03-22 17:37:07]

Turns out that the Vista failure was ill-timed flake.  I saw the bot fail
similarly yesterday, so I tried recommitting this patch and it stuck without
issue.