Add scheme to HostZoomMap

Web Content zoom settings should not apply to login/lock screen

Add scheme to HostZoomMap, use it to specify unmodifiable zoom level for internal ui pages.

Reason : Currently, zoom settings are remebered on a per-host basis, and ignore scheme.
It works nice for http/https access for the same host, but it is not possible to distinguish between
internal ui (with chrome:// scheme) and some intranet sites (having https://login/ does not seem unlikely).

BUG=164269
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=187331

[Nikita (slow), 2013-01-11 15:23:55]

https://chromiumcodereview.appspot.com/11866004/diff/1/chrome/browser/profile...
File chrome/browser/profiles/profile_impl.cc (right):

https://chromiumcodereview.appspot.com/11866004/diff/1/chrome/browser/profile...
chrome/browser/profiles/profile_impl.cc:562: chrome::kChromeUIOobeHost,
nit: Alignment like this is more readable if it fits

host_zoom_map->SetZoomLevel(chrome::kChromeUIUberHost,
                            chrome::kChromeUILoginHost,
                            0);

https://chromiumcodereview.appspot.com/11866004/diff/1/content/public/browser...
File content/public/browser/host_zoom_map.h (right):

https://chromiumcodereview.appspot.com/11866004/diff/1/content/public/browser...
content/public/browser/host_zoom_map.h:42: // Returns the zoom level for the
host or spec for a given url, taking scheme
nit: |scheme| |host|

https://chromiumcodereview.appspot.com/11866004/diff/1/content/public/browser...
content/public/browser/host_zoom_map.h:48: virtual double GetZoomLevel(const
std::string& scheme,
Maybe name this GetZoomLevelBySchemeHost ?

http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Function_Overl...

https://chromiumcodereview.appspot.com/11866004/diff/1/content/renderer/rende...
File content/renderer/render_thread_impl.cc (right):

https://chromiumcodereview.appspot.com/11866004/diff/1/content/renderer/rende...
content/renderer/render_thread_impl.cc:160: (scheme_.empty() || scheme_ ==
url.scheme()))
nit: move one space left

https://chromiumcodereview.appspot.com/11866004/diff/1/content/renderer/rende...
content/renderer/render_thread_impl.cc:161: webview->setZoomLevel(false,
zoom_level_);
nit: add {}

[Denis Kuznetsov (DE-MUC), 2013-01-11 15:25:11]

Dave for chrome/browser/profiles/profile_impl.cc
Scott for  content/browser/
Jói for content/public/
James for content/renderer/
Cris for content/common/
Nikita for quick local overeview.

[Denis Kuznetsov (DE-MUC), 2013-01-11 15:48:59]

https://chromiumcodereview.appspot.com/11866004/diff/1/content/public/browser...
File content/public/browser/host_zoom_map.h (right):

https://chromiumcodereview.appspot.com/11866004/diff/1/content/public/browser...
content/public/browser/host_zoom_map.h:42: // Returns the zoom level for the
host or spec for a given url, taking scheme
On 2013/01/11 15:23:55, Nikita Kostylev wrote:
> nit: |scheme| |host|

Done.

https://chromiumcodereview.appspot.com/11866004/diff/1/content/public/browser...
content/public/browser/host_zoom_map.h:48: virtual double GetZoomLevel(const
std::string& scheme,
On 2013/01/11 15:23:55, Nikita Kostylev wrote:
> Maybe name this GetZoomLevelBySchemeHost ?
> 
>
http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Function_Overl...
I would prefer to rename or remove old method, as it is now useless, however,
that would be much more incompatible API change.

https://chromiumcodereview.appspot.com/11866004/diff/1/content/renderer/rende...
File content/renderer/render_thread_impl.cc (right):

https://chromiumcodereview.appspot.com/11866004/diff/1/content/renderer/rende...
content/renderer/render_thread_impl.cc:160: (scheme_.empty() || scheme_ ==
url.scheme()))
On 2013/01/11 15:23:55, Nikita Kostylev wrote:
> nit: move one space left

Done.

https://chromiumcodereview.appspot.com/11866004/diff/1/content/renderer/rende...
content/renderer/render_thread_impl.cc:161: webview->setZoomLevel(false,
zoom_level_);
On 2013/01/11 15:23:55, Nikita Kostylev wrote:
> nit: add {}

Done.

[Nikita (slow), 2013-01-11 15:50:39]

https://chromiumcodereview.appspot.com/11866004/diff/1/content/public/browser...
File content/public/browser/host_zoom_map.h (right):

https://chromiumcodereview.appspot.com/11866004/diff/1/content/public/browser...
content/public/browser/host_zoom_map.h:48: virtual double GetZoomLevel(const
std::string& scheme,
On 2013/01/11 15:48:59, Denis Kuznetsov wrote:
> On 2013/01/11 15:23:55, Nikita Kostylev wrote:
> > Maybe name this GetZoomLevelBySchemeHost ?
> > 
> >
>
http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Function_Overl...
> I would prefer to rename or remove old method, as it is now useless, however,
> that would be much more incompatible API change.

I agree that this seems like a better option.

[Jói, 2013-01-11 17:41:09]

LGTM with nit.

https://chromiumcodereview.appspot.com/11866004/diff/5001/content/browser/hos...
File content/browser/host_zoom_map_impl.h (right):

https://chromiumcodereview.appspot.com/11866004/diff/5001/content/browser/hos...
content/browser/host_zoom_map_impl.h:35: virtual double GetZoomLevel(const
std::string& scheme,
Order these the same as in HostZoomMap.

[Cris Neckar, 2013-01-11 19:38:02]

IPC security audit LGTM

[sky, 2013-01-11 20:33:32]

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
File content/browser/host_zoom_map_impl.cc (right):

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
content/browser/host_zoom_map_impl.cc:50: for (HostZoomLevels::const_iterator
i(source.begin());
dest->insert(source.begin(), source.end()) , at which point you might want to
nuke the function.

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
content/browser/host_zoom_map_impl.cc:150: 
remove one newline.

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
File content/browser/host_zoom_map_impl.h (right):

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
content/browser/host_zoom_map_impl.h:89: void CopyHostZoomLevels(const
HostZoomLevels &source,
& close to type, and HostZoomLevels should be a * (from style guide: "while
output and input/output parameters will be non-const pointers").

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
content/browser/host_zoom_map_impl.h:92: 
nit: only one newline here.

[jamesr, 2013-01-12 00:38:31]

content/renderer lgtm

[Nikita (slow), 2013-01-16 15:53:17]

friendly ping

[Nikita (slow), 2013-01-16 15:54:10]

Dave, Scott friendly ping

[sky, 2013-01-16 17:03:32]

AFAICT the patch has not been updated since last comments.

[Denis Kuznetsov (DE-MUC), 2013-01-17 13:23:47]

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
File content/browser/host_zoom_map_impl.cc (right):

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
content/browser/host_zoom_map_impl.cc:50: for (HostZoomLevels::const_iterator
i(source.begin());
On 2013/01/11 20:33:32, sky wrote:
> dest->insert(source.begin(), source.end()) , at which point you might want to
> nuke the function.

Done.

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
content/browser/host_zoom_map_impl.cc:150: 
On 2013/01/11 20:33:32, sky wrote:
> remove one newline.

Done.

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
content/browser/host_zoom_map_impl.cc:150: 
On 2013/01/11 20:33:32, sky wrote:
> remove one newline.

Done.

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
File content/browser/host_zoom_map_impl.h (right):

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
content/browser/host_zoom_map_impl.h:35: virtual double GetZoomLevel(const
std::string& scheme,
On 2013/01/11 17:41:09, Jói wrote:
> Order these the same as in HostZoomMap.

Done.

https://codereview.chromium.org/11866004/diff/5001/content/browser/host_zoom_...
content/browser/host_zoom_map_impl.h:92: 
On 2013/01/11 20:33:32, sky wrote:
> nit: only one newline here.

Done.

[sky, 2013-01-17 17:08:01]

https://codereview.chromium.org/11866004/diff/20001/content/browser/host_zoom...
File content/browser/host_zoom_map_impl.h (right):

https://codereview.chromium.org/11866004/diff/20001/content/browser/host_zoom...
content/browser/host_zoom_map_impl.h:37: virtual void SetZoomLevel(const
std::string& scheme,
Do we really need both of these (same comment for the getter). Wouldn't it be
cleaner to have only one method that takes the scheme and host?

[Denis Kuznetsov (DE-MUC), 2013-01-21 10:49:29]

https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
File content/browser/host_zoom_map_impl.h (right):

https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
content/browser/host_zoom_map_impl.h:37: virtual void SetZoomLevel(const
std::string& scheme,
On 2013/01/17 17:08:01, sky wrote:
> Do we really need both of these (same comment for the getter). Wouldn't it be
> cleaner to have only one method that takes the scheme and host?

Actually, the cleanest way to do this is to make HostZoomMap  accept GURL both
for Get and Set zoom levels, and let HostZoomMap to handle zoom-equivalence
classes of URLs (e.g. for http:// and https:// it is the URLs with same domain,
for internal settings pages that is all pages with chrome:// scheme, except for
/login/ and /oobe/ parts, etc).

However, as this is the part of content-api, I've tried to extend API, but leave
it backwards-compatible.

[sky, 2013-01-22 16:35:34]

On Mon, Jan 21, 2013 at 2:49 AM,  <[email protected]> wrote:
>
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
> File content/browser/host_zoom_map_impl.h (right):
>
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
>
> content/browser/host_zoom_map_impl.h:37: virtual void SetZoomLevel(const
> std::string& scheme,
> On 2013/01/17 17:08:01, sky wrote:
>>
>> Do we really need both of these (same comment for the getter).
>
> Wouldn't it be
>>
>> cleaner to have only one method that takes the scheme and host?
>
>
> Actually, the cleanest way to do this is to make HostZoomMap  accept
> GURL both for Get and Set zoom levels, and let HostZoomMap to handle
> zoom-equivalence classes of URLs (e.g. for http:// and https:// it is
> the URLs with same domain, for internal settings pages that is all pages
> with chrome:// scheme, except for /login/ and /oobe/ parts, etc).

Ya, I like that.

> However, as this is the part of content-api, I've tried to extend API,
> but leave it backwards-compatible.

Huh? Why can't you change content-api?

  -Scott

[DaveMoore, 2013-01-29 01:15:58]

lgtm

[Denis Kuznetsov (DE-MUC), 2013-02-04 19:11:05]

On 2013/01/22 16:35:34, sky wrote:
> On Mon, Jan 21, 2013 at 2:49 AM,  <mailto:[email protected]> wrote:
> >
> >
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
> > File content/browser/host_zoom_map_impl.h (right):
> >
> >
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
> >
> > content/browser/host_zoom_map_impl.h:37: virtual void SetZoomLevel(const
> > std::string& scheme,
> > On 2013/01/17 17:08:01, sky wrote:
> >>
> >> Do we really need both of these (same comment for the getter).
> >
> > Wouldn't it be
> >>
> >> cleaner to have only one method that takes the scheme and host?
> >
> >
> > Actually, the cleanest way to do this is to make HostZoomMap  accept
> > GURL both for Get and Set zoom levels, and let HostZoomMap to handle
> > zoom-equivalence classes of URLs (e.g. for http:// and https:// it is
> > the URLs with same domain, for internal settings pages that is all pages
> > with chrome:// scheme, except for /login/ and /oobe/ parts, etc).
> 
> Ya, I like that.

Unfortunatelly, that does not work as easy as I thought - in a renderer we need
to check if the URI renderer has matches host/scheme in update message, and that
would require having introducing "matcher"  to renderer.

I've just removed old methods.

[sky, 2013-02-05 16:57:49]

On Mon, Feb 4, 2013 at 11:11 AM,  <[email protected]> wrote:
> On 2013/01/22 16:35:34, sky wrote:
>
>> On Mon, Jan 21, 2013 at 2:49 AM,  <mailto:[email protected]> wrote:
>> >
>> >
>
>
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
>>
>> > File content/browser/host_zoom_map_impl.h (right):
>> >
>> >
>
>
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
>>
>> >
>> > content/browser/host_zoom_map_impl.h:37: virtual void SetZoomLevel(const
>> > std::string& scheme,
>> > On 2013/01/17 17:08:01, sky wrote:
>> >>
>> >> Do we really need both of these (same comment for the getter).
>> >
>> > Wouldn't it be
>> >>
>> >> cleaner to have only one method that takes the scheme and host?
>> >
>> >
>> > Actually, the cleanest way to do this is to make HostZoomMap  accept
>> > GURL both for Get and Set zoom levels, and let HostZoomMap to handle
>> > zoom-equivalence classes of URLs (e.g. for http:// and https:// it is
>> > the URLs with same domain, for internal settings pages that is all pages
>> > with chrome:// scheme, except for /login/ and /oobe/ parts, etc).
>
>
>> Ya, I like that.
>
>
> Unfortunatelly, that does not work as easy as I thought - in a renderer we
> need
> to check if the URI renderer has matches host/scheme in update message, and
> that
> would require having introducing "matcher"  to renderer.

What specific place are you referring to?

[Denis Kuznetsov (DE-MUC), 2013-02-05 17:14:25]

On 2013/02/05 16:57:49, sky wrote:
> On Mon, Feb 4, 2013 at 11:11 AM,  <mailto:[email protected]> wrote:
> > On 2013/01/22 16:35:34, sky wrote:
> >
> >> On Mon, Jan 21, 2013 at 2:49 AM,  <mailto:[email protected]> wrote:
> >> >
> >> >
> >
> >
> >
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
> >>
> >> > File content/browser/host_zoom_map_impl.h (right):
> >> >
> >> >
> >
> >
> >
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
> >>
> >> >
> >> > content/browser/host_zoom_map_impl.h:37: virtual void SetZoomLevel(const
> >> > std::string& scheme,
> >> > On 2013/01/17 17:08:01, sky wrote:
> >> >>
> >> >> Do we really need both of these (same comment for the getter).
> >> >
> >> > Wouldn't it be
> >> >>
> >> >> cleaner to have only one method that takes the scheme and host?
> >> >
> >> >
> >> > Actually, the cleanest way to do this is to make HostZoomMap  accept
> >> > GURL both for Get and Set zoom levels, and let HostZoomMap to handle
> >> > zoom-equivalence classes of URLs (e.g. for http:// and https:// it is
> >> > the URLs with same domain, for internal settings pages that is all pages
> >> > with chrome:// scheme, except for /login/ and /oobe/ parts, etc).
> >
> >
> >> Ya, I like that.
> >
> >
> > Unfortunatelly, that does not work as easy as I thought - in a renderer we
> > need
> > to check if the URI renderer has matches host/scheme in update message, and
> > that
> > would require having introducing "matcher"  to renderer.
> 
> What specific place are you referring to?

If we change code so that HostZoomMap hides all the details about which parts of
GURL determine same "zoom domain", then we would need some replacement for
updating all pages within this "zoom domain" once zoom value changes.
Currently it is done via ViewMsg_SetZoomLevelForCurrentURL message to
renderer_thread, and RenderViewZoomer checks if document.url() matches
host/host+scheme in message. 
But in possible implementation above we would need some special object that try
to match renderer's document.url() to check if zoom level have to be updated
within renderer.

[sky, 2013-02-05 20:44:54]

On Tue, Feb 5, 2013 at 9:14 AM,  <[email protected]> wrote:
> On 2013/02/05 16:57:49, sky wrote:
>
>> On Mon, Feb 4, 2013 at 11:11 AM,  <mailto:[email protected]> wrote:
>> > On 2013/01/22 16:35:34, sky wrote:
>> >
>> >> On Mon, Jan 21, 2013 at 2:49 AM,  <mailto:[email protected]> wrote:
>> >> >
>> >> >
>> >
>> >
>> >
>
>
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
>>
>> >>
>> >> > File content/browser/host_zoom_map_impl.h (right):
>> >> >
>> >> >
>> >
>> >
>> >
>
>
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
>>
>> >>
>> >> >
>> >> > content/browser/host_zoom_map_impl.h:37: virtual void
>> >> > SetZoomLevel(const
>> >> > std::string& scheme,
>> >> > On 2013/01/17 17:08:01, sky wrote:
>> >> >>
>> >> >> Do we really need both of these (same comment for the getter).
>> >> >
>> >> > Wouldn't it be
>> >> >>
>> >> >> cleaner to have only one method that takes the scheme and host?
>> >> >
>> >> >
>> >> > Actually, the cleanest way to do this is to make HostZoomMap  accept
>> >> > GURL both for Get and Set zoom levels, and let HostZoomMap to handle
>> >> > zoom-equivalence classes of URLs (e.g. for http:// and https:// it is
>> >> > the URLs with same domain, for internal settings pages that is all
>> >> > pages
>> >> > with chrome:// scheme, except for /login/ and /oobe/ parts, etc).
>> >
>> >
>> >> Ya, I like that.
>> >
>> >
>> > Unfortunatelly, that does not work as easy as I thought - in a renderer
>> > we
>> > need
>> > to check if the URI renderer has matches host/scheme in update message,
>> > and
>> > that
>> > would require having introducing "matcher"  to renderer.
>
>
>> What specific place are you referring to?
>
>
> If we change code so that HostZoomMap hides all the details about which
> parts of
> GURL determine same "zoom domain", then we would need some replacement for
> updating all pages within this "zoom domain" once zoom value changes.
> Currently it is done via ViewMsg_SetZoomLevelForCurrentURL message to
> renderer_thread, and RenderViewZoomer checks if document.url() matches
> host/host+scheme in message.
> But in possible implementation above we would need some special object that
> try
> to match renderer's document.url() to check if zoom level have to be updated
> within renderer.
>
> https://chromiumcodereview.appspot.com/11866004/

Are you saying we would need something like
SetZoomLevelForAllSchemes(GURL url)? Isn't that clearer than having an
empty scheme be interpreted differently?

  -Scott

[Denis Kuznetsov (DE-MUC), 2013-02-07 17:03:06]

On 2013/02/05 20:44:54, sky wrote:
> On Tue, Feb 5, 2013 at 9:14 AM,  <mailto:[email protected]> wrote:
> > On 2013/02/05 16:57:49, sky wrote:
> >
> >> On Mon, Feb 4, 2013 at 11:11 AM,  <mailto:[email protected]> wrote:
> >> > On 2013/01/22 16:35:34, sky wrote:
> >> >
> >> >> On Mon, Jan 21, 2013 at 2:49 AM,  <mailto:[email protected]> wrote:
> >> >> >
> >> >> >
> >> >
> >> >
> >> >
> >
> >
> >
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
> >>
> >> >>
> >> >> > File content/browser/host_zoom_map_impl.h (right):
> >> >> >
> >> >> >
> >> >
> >> >
> >> >
> >
> >
> >
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
> >>
> >> >>
> >> >> >
> >> >> > content/browser/host_zoom_map_impl.h:37: virtual void
> >> >> > SetZoomLevel(const
> >> >> > std::string& scheme,
> >> >> > On 2013/01/17 17:08:01, sky wrote:
> >> >> >>
> >> >> >> Do we really need both of these (same comment for the getter).
> >> >> >
> >> >> > Wouldn't it be
> >> >> >>
> >> >> >> cleaner to have only one method that takes the scheme and host?
> >> >> >
> >> >> >
> >> >> > Actually, the cleanest way to do this is to make HostZoomMap  accept
> >> >> > GURL both for Get and Set zoom levels, and let HostZoomMap to handle
> >> >> > zoom-equivalence classes of URLs (e.g. for http:// and https:// it is
> >> >> > the URLs with same domain, for internal settings pages that is all
> >> >> > pages
> >> >> > with chrome:// scheme, except for /login/ and /oobe/ parts, etc).
> >> >
> >> >
> >> >> Ya, I like that.
> >> >
> >> >
> >> > Unfortunatelly, that does not work as easy as I thought - in a renderer
> >> > we
> >> > need
> >> > to check if the URI renderer has matches host/scheme in update message,
> >> > and
> >> > that
> >> > would require having introducing "matcher"  to renderer.
> >
> >
> >> What specific place are you referring to?
> >
> >
> > If we change code so that HostZoomMap hides all the details about which
> > parts of
> > GURL determine same "zoom domain", then we would need some replacement for
> > updating all pages within this "zoom domain" once zoom value changes.
> > Currently it is done via ViewMsg_SetZoomLevelForCurrentURL message to
> > renderer_thread, and RenderViewZoomer checks if document.url() matches
> > host/host+scheme in message.
> > But in possible implementation above we would need some special object that
> > try
> > to match renderer's document.url() to check if zoom level have to be updated
> > within renderer.
> >
> > https://chromiumcodereview.appspot.com/11866004/
> 
> Are you saying we would need something like
> SetZoomLevelForAllSchemes(GURL url)? Isn't that clearer than having an
> empty scheme be interpreted differently?
> 
>   -Scott
I'm saying that best API for HostZoomMap should be SetZoom(GURL, Zoom) /
GetZoom(GURL).
That would make it possible to have zoom levels for same host with different
schemas, for hostname.com/www.hostname.com.
It would also be possible to have same zoom levels for completely different
GURLs, e.g. now we have chrome://settings and chrome://history, and they have
same zoom level only because internally these URLs are transformed to
chrome://chrome/settings and  chrome://chrome/history, and, thus, have same
domain (if this url substitution is changed some day, there will be zooming
issues).
But implementation that allow this would need some object that would have either
method 
GURL GetCanonicalURLForZoom(GURL) or 
bool HaveSameZoom(GURL, GURL), and
this object should be available in Renderer.

[sky, 2013-02-11 17:13:16]

I could be wrong here, but isn't the renderer told about the zoom
level by way of:

    filter_->Send(new ViewMsg_SetZoomLevelForLoadingURL(
        info->GetRouteID(),
        request_url, host_zoom_map->GetZoomLevel(net::GetHostOrSpecFromURL(
            request_url))));

(from async_resource_handler.cc) which has the url?

  -Scott

On Thu, Feb 7, 2013 at 9:03 AM,  <[email protected]> wrote:
> On 2013/02/05 20:44:54, sky wrote:
>
>> On Tue, Feb 5, 2013 at 9:14 AM,  <mailto:[email protected]> wrote:
>> > On 2013/02/05 16:57:49, sky wrote:
>> >
>> >> On Mon, Feb 4, 2013 at 11:11 AM,  <mailto:[email protected]> wrote:
>> >> > On 2013/01/22 16:35:34, sky wrote:
>> >> >
>> >> >> On Mon, Jan 21, 2013 at 2:49 AM,  <mailto:[email protected]>
>> >> >> wrote:
>> >> >> >
>> >> >> >
>> >> >
>> >> >
>> >> >
>> >
>> >
>> >
>
>
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
>>
>> >>
>> >> >>
>> >> >> > File content/browser/host_zoom_map_impl.h (right):
>> >> >> >
>> >> >> >
>> >> >
>> >> >
>> >> >
>> >
>> >
>> >
>
>
>
https://chromiumcodereview.appspot.com/11866004/diff/20001/content/browser/ho...
>>
>> >>
>> >> >>
>> >> >> >
>> >> >> > content/browser/host_zoom_map_impl.h:37: virtual void
>> >> >> > SetZoomLevel(const
>> >> >> > std::string& scheme,
>> >> >> > On 2013/01/17 17:08:01, sky wrote:
>> >> >> >>
>> >> >> >> Do we really need both of these (same comment for the getter).
>> >> >> >
>> >> >> > Wouldn't it be
>> >> >> >>
>> >> >> >> cleaner to have only one method that takes the scheme and host?
>> >> >> >
>> >> >> >
>> >> >> > Actually, the cleanest way to do this is to make HostZoomMap
>> >> >> > accept
>> >> >> > GURL both for Get and Set zoom levels, and let HostZoomMap to
>> >> >> > handle
>> >> >> > zoom-equivalence classes of URLs (e.g. for http:// and https:// it
>> >> >> > is
>> >> >> > the URLs with same domain, for internal settings pages that is all
>> >> >> > pages
>> >> >> > with chrome:// scheme, except for /login/ and /oobe/ parts, etc).
>> >> >
>> >> >
>> >> >> Ya, I like that.
>> >> >
>> >> >
>> >> > Unfortunatelly, that does not work as easy as I thought - in a
>> >> > renderer
>> >> > we
>> >> > need
>> >> > to check if the URI renderer has matches host/scheme in update
>> >> > message,
>> >> > and
>> >> > that
>> >> > would require having introducing "matcher"  to renderer.
>> >
>> >
>> >> What specific place are you referring to?
>> >
>> >
>> > If we change code so that HostZoomMap hides all the details about which
>> > parts of
>> > GURL determine same "zoom domain", then we would need some replacement
>> > for
>> > updating all pages within this "zoom domain" once zoom value changes.
>> > Currently it is done via ViewMsg_SetZoomLevelForCurrentURL message to
>> > renderer_thread, and RenderViewZoomer checks if document.url() matches
>> > host/host+scheme in message.
>> > But in possible implementation above we would need some special object
>> > that
>> > try
>> > to match renderer's document.url() to check if zoom level have to be
>> > updated
>> > within renderer.
>> >
>> > https://chromiumcodereview.appspot.com/11866004/
>
>
>> Are you saying we would need something like
>> SetZoomLevelForAllSchemes(GURL url)? Isn't that clearer than having an
>> empty scheme be interpreted differently?
>
>
>>    -Scott
>
> I'm saying that best API for HostZoomMap should be SetZoom(GURL, Zoom) /
> GetZoom(GURL).
> That would make it possible to have zoom levels for same host with different
> schemas, for hostname.com/www.hostname.com.
> It would also be possible to have same zoom levels for completely different
> GURLs, e.g. now we have chrome://settings and chrome://history, and they
> have
> same zoom level only because internally these URLs are transformed to
> chrome://chrome/settings and  chrome://chrome/history, and, thus, have same
> domain (if this url substitution is changed some day, there will be zooming
> issues).
> But implementation that allow this would need some object that would have
> either
> method
> GURL GetCanonicalURLForZoom(GURL) or
> bool HaveSameZoom(GURL, GURL), and
> this object should be available in Renderer.
>
> https://chromiumcodereview.appspot.com/11866004/

[sky, 2013-02-11 17:13:26]

Here's a couple of other nits.

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/browser/ho...
File content/browser/host_zoom_map_impl_unittest.cc (right):

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/browser/ho...
content/browser/host_zoom_map_impl_unittest.cc:46:
EXPECT_DOUBLE_EQ(host_zoom_map.GetZoomLevel("chrome", "login"), 0);
arguments to expect/assert should be expected then actual.

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/browser/re...
File content/browser/renderer_host/render_view_host_impl.cc (right):

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/browser/re...
content/browser/renderer_host/render_view_host_impl.cc:1927: std::string(),
net::GetHostOrSpecFromURL(url), zoom_level);
Why don't you use the scheme from |url|?

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
File content/renderer/render_thread_impl.cc (right):

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
content/renderer/render_thread_impl.cc:145: const std::string& host,
align with '(' on previous line (same comment for 146).

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
content/renderer/render_thread_impl.cc:147: : scheme_(scheme), host_(host),
zoom_level_(zoom_level) {
since the constructor doesn't fit on one line each parameter should be on its
own line.

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
content/renderer/render_thread_impl.cc:160: (scheme_.empty() || scheme_ ==
url.scheme())) {
Why the empty() check?

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
content/renderer/render_thread_impl.cc:167: std::string scheme_;
const on these.

[sky, 2013-02-11 19:20:31]

Denis pointed out that zoom level applies to the pair scheme+host, so that using
a GURL here is misleading.

Joi, currently the patch treats an empty string as applying to all schemes,
which is a bit mysterious. What do you think of two explicit setters:

SetZoomLevelForHost(host, zoom);
SetZoomLevelForHostAndSheme(scheme, host, zoom);

The former applies to all schemes, the later only the specified scheme.

[Jói, 2013-02-11 19:42:36]

> Joi, currently the patch treats an empty string as applying to all schemes,
> which is a bit mysterious. What do you think of two explicit setters:
> 
> SetZoomLevelForHost(host, zoom);
> SetZoomLevelForHostAndSheme(scheme, host, zoom);

SGTM


On 2013/02/11 19:20:31, sky wrote:
> Denis pointed out that zoom level applies to the pair scheme+host, so that
using
> a GURL here is misleading.
> 
> Joi, currently the patch treats an empty string as applying to all schemes,
> which is a bit mysterious. What do you think of two explicit setters:
> 
> SetZoomLevelForHost(host, zoom);
> SetZoomLevelForHostAndSheme(scheme, host, zoom);
> 
> The former applies to all schemes, the later only the specified scheme.

[Denis Kuznetsov (DE-MUC), 2013-02-12 13:10:45]

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/browser/ho...
File content/browser/host_zoom_map_impl_unittest.cc (right):

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/browser/ho...
content/browser/host_zoom_map_impl_unittest.cc:46:
EXPECT_DOUBLE_EQ(host_zoom_map.GetZoomLevel("chrome", "login"), 0);
On 2013/02/11 17:13:26, sky wrote:
> arguments to expect/assert should be expected then actual.

Done.

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
File content/renderer/render_thread_impl.cc (right):

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
content/renderer/render_thread_impl.cc:145: const std::string& host,
On 2013/02/11 17:13:26, sky wrote:
> align with '(' on previous line (same comment for 146).

Done.

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
content/renderer/render_thread_impl.cc:147: : scheme_(scheme), host_(host),
zoom_level_(zoom_level) {
On 2013/02/11 17:13:26, sky wrote:
> since the constructor doesn't fit on one line each parameter should be on its
> own line.

Done.

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
content/renderer/render_thread_impl.cc:160: (scheme_.empty() || scheme_ ==
url.scheme())) {
On 2013/02/11 17:13:26, sky wrote:
> Why the empty() check?
We use same message for SetForHost and SetForHostAndScheme, empty scheme is a
wildcard that matches any scheme.

https://chromiumcodereview.appspot.com/11866004/diff/26001/content/renderer/r...
content/renderer/render_thread_impl.cc:167: std::string scheme_;
On 2013/02/11 17:13:26, sky wrote:
> const on these.

Done.

[sky, 2013-02-12 18:47:14]

https://chromiumcodereview.appspot.com/11866004/diff/37002/content/public/bro...
File content/public/browser/host_zoom_map.h (right):

https://chromiumcodereview.appspot.com/11866004/diff/37002/content/public/bro...
content/public/browser/host_zoom_map.h:41: virtual double
GetZoomLevelForHost(const std::string& host) const = 0;
How can we can keep this exposed as public? Doing so makes it easy to circumvent
the zoom set for a scheme+host pair. For example, if the zoom is changed for a
particular scheme+host, the callbacks are notified and one of the callbacks ask
for the zoom level for the host they may get the wrong thing. I think you need
to nuke this and make the callbacks get the scheme+host pair that changed.

https://chromiumcodereview.appspot.com/11866004/diff/37002/content/public/bro...
content/public/browser/host_zoom_map.h:43: // Returns the zoom level for the
host or spec for a given url, taking
Update description for both this and GetZoomLevelForHost. For example, there is
no 'spec' here or above. Similarly there is no url here or above.

[Denis Kuznetsov (DE-MUC), 2013-03-06 19:05:25]

Here is new version as we discussed - GetZoomLevelForHost is not in API anymore,
and callbacks got more complex.

[Jói, 2013-03-06 20:43:17]

content/public: Couple of documentation things to address.

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
File content/public/browser/host_zoom_map.h (right):

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:29: // Enum that indicates how
trailing sentence

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:56: // level is determined by the host
portion of the URL, or (in the absence of
"in the absence of a host" seems a bit unclear to me. Does it mean the absence
of a host in the |host| parameter, or does it mean the absence of a registration
for that particular host, when a registration for the full spec passed as |host|
does exist?

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:74: // Sets the zoom level for the
|scheme|/|host| pair to |level|.
One point you might clarify in the documentation for this; |host| in this case
can still be a host or a spec, is that right?

[sky, 2013-03-06 22:21:11]

Looking much better, than you.

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
File content/public/browser/host_zoom_map.h (right):

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:33: ZOOM_CHANGED_GENERIC,            //
Some zoom level change happened,
It is totally unclear what this is, and because of that it isn't clear what a
callback should do. This seems to only be invoked when the temporary zoom level
changes. Can we name this to match: ZOOM_CHANGED_TEMPORARY_ZOOM with a suitable
comment (and indicate host/scheme are not set).

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:72: virtual void
SetZoomLevelForHost(const std::string& host, double level) = 0;
Can you document at the class level what the precedence is for these? By that I
mean things like setting the zoom level for a host does not override the zoom
level for scheme-host.

[Denis Kuznetsov (DE-MUC), 2013-03-07 15:16:59]

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
File content/public/browser/host_zoom_map.h (right):

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:29: // Enum that indicates how
On 2013/03/06 20:43:17, Jói wrote:
> trailing sentence

Done.

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:33: ZOOM_CHANGED_GENERIC,            //
Some zoom level change happened,
On 2013/03/06 22:21:11, sky wrote:
> It is totally unclear what this is, and because of that it isn't clear what a
> callback should do. This seems to only be invoked when the temporary zoom
level
> changes. Can we name this to match: ZOOM_CHANGED_TEMPORARY_ZOOM with a
suitable
> comment (and indicate host/scheme are not set).

Done.

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:56: // level is determined by the host
portion of the URL, or (in the absence of
On 2013/03/06 20:43:17, Jói wrote:
> "in the absence of a host" seems a bit unclear to me. Does it mean the absence
> of a host in the |host| parameter, or does it mean the absence of a
registration
> for that particular host, when a registration for the full spec passed as
|host|
> does exist?

Done.

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:72: virtual void
SetZoomLevelForHost(const std::string& host, double level) = 0;
On 2013/03/06 22:21:11, sky wrote:
> Can you document at the class level what the precedence is for these? By that
I
> mean things like setting the zoom level for a host does not override the zoom
> level for scheme-host.

Done.

https://codereview.chromium.org/11866004/diff/50001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:74: // Sets the zoom level for the
|scheme|/|host| pair to |level|.
On 2013/03/06 20:43:17, Jói wrote:
> One point you might clarify in the documentation for this; |host| in this case
> can still be a host or a spec, is that right?

Done.

[sky, 2013-03-07 15:47:02]

LGTM

https://codereview.chromium.org/11866004/diff/54001/chrome/browser/profiles/o...
File chrome/browser/profiles/off_the_record_profile_impl.cc (right):

https://codereview.chromium.org/11866004/diff/54001/chrome/browser/profiles/o...
chrome/browser/profiles/off_the_record_profile_impl.cc:489: default:
Remove default so that we get compile time error if new constant is added.

https://codereview.chromium.org/11866004/diff/54001/content/public/browser/ho...
File content/public/browser/host_zoom_map.h (right):

https://codereview.chromium.org/11866004/diff/54001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:27: // setting
For that last sentence, how about: "Getting the zoom level starts at the most
specific setting and progresses to the less specific: first the zoom for the
host and scheme pair is checked, secondly the zoom for the host only and lastly
default zoom."

https://codereview.chromium.org/11866004/diff/54001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:40: // Structure used to notify about
zoom changes. Host and Scheme part can be
Host and/or scheme are empty if not applicable to |mode|.

https://codereview.chromium.org/11866004/diff/54001/content/public/browser/ho...
content/public/browser/host_zoom_map.h:58: // Returns the zoom level for the
|host| for a given url, taking |scheme| in
I think I nuke all this and go with:
// Reeturns the zoom for the specified scheme and host. See class description
for details.

[commit-bot: I haz the power, 2013-03-07 16:28:36]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11866004/60001

[commit-bot: I haz the power, 2013-03-07 16:36:11]

Failed to trigger a try job on win_x64_rel
HTTP Error 400: Bad Request

[commit-bot: I haz the power, 2013-03-07 16:36:16]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11866004/72001

[commit-bot: I haz the power, 2013-03-07 16:59:31]

Sorry for I got bad news for ya.
Compile failed with a clobber build on linux_rel.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_rel&...
Your code is likely broken or HEAD is junk. Please ensure your
code is not broken then alert the build sheriffs.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2013-03-07 18:35:24]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11866004/75001

[commit-bot: I haz the power, 2013-03-07 19:15:55]

Sorry for I got bad news for ya.
Compile failed with a clobber build on linux_rel.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_rel&...
Your code is likely broken or HEAD is junk. Please ensure your
code is not broken then alert the build sheriffs.
Look at the try server FAQ for more details.

[commit-bot: I haz the power, 2013-03-11 12:23:39]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11866004/83002

[commit-bot: I haz the power, 2013-03-11 14:32:05]

Retried try job too often on win_rel for step(s) browser_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2013-03-11 14:47:18]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11866004/83002

[commit-bot: I haz the power, 2013-03-11 18:59:05]

Change committed as 187331