Remove --testing-fixed-http-port and --testing-fixed-https-port.

Remove --testing-fixed-http-port and --testing-fixed-https-port.

--host-rules seems like a reasonable enough substitute for these two command line flags, and removing them will simplifies the fairly complicated URL rewrite logic in HttpNetworkTransaction.

BUG=None
TEST=built and ran net_unittests

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=53790

[cbentzel, 2010-07-26 13:21:55]

The only concern with dropping these two flags is that --host-rules does rules
matching using the host, but never the port.

If you are testing against a single server name, it doesn't seem to onerous to
add the host name for matching to the new port. If multiple (possibly in the
case of the SPDY in memory server?) I'll need to add host matches based on
destination port before making this change.

[mbelshe, 2010-07-26 16:42:40]

I do use these every now and then...

Mike


On Mon, Jul 26, 2010 at 6:21 AM, <cbentzel@chromium.org> wrote:

> Reviewers: Mike Belshe, eroman,
>
> Message:
> The only concern with dropping these two flags is that --host-rules does
> rules
> matching using the host, but never the port.
>
> If you are testing against a single server name, it doesn't seem to onerous
> to
> add the host name for matching to the new port. If multiple (possibly in
> the
> case of the SPDY in memory server?) I'll need to add host matches based on
> destination port before making this change.
>
> Description:
> Remove --testing-fixed-http-port and --testing-fixed-https-port.
>
> --host-rules seems like a reasonable enough substitute for these two
> command
> line flags, and removing them will simplifies the fairly complicated URL
> rewrite
> logic in HttpNetworkTransaction.
>
> BUG=None
> TEST=built and ran net_unittests
>
> Please review this at http://codereview.chromium.org/3041019/show
>
> Affected files:
>  M chrome/browser/browser_main.cc
>  M chrome/common/chrome_switches.h
>  M chrome/common/chrome_switches.cc
>  M net/http/http_network_session.h
>  M net/http/http_network_session.cc
>  M net/http/http_network_transaction.cc
>  M net/spdy/spdy_network_transaction.cc
>
>
>

[cbentzel, 2010-07-26 16:54:50]

Would host-rules be sufficient for you?

On Mon, Jul 26, 2010 at 12:42 PM, Mike Belshe <mbelshe@google.com> wrote:

> I do use these every now and then...
>
> Mike
>
>
> On Mon, Jul 26, 2010 at 6:21 AM, <cbentzel@chromium.org> wrote:
>
>> Reviewers: Mike Belshe, eroman,
>>
>> Message:
>> The only concern with dropping these two flags is that --host-rules does
>> rules
>> matching using the host, but never the port.
>>
>> If you are testing against a single server name, it doesn't seem to
>> onerous to
>> add the host name for matching to the new port. If multiple (possibly in
>> the
>> case of the SPDY in memory server?) I'll need to add host matches based on
>> destination port before making this change.
>>
>> Description:
>> Remove --testing-fixed-http-port and --testing-fixed-https-port.
>>
>> --host-rules seems like a reasonable enough substitute for these two
>> command
>> line flags, and removing them will simplifies the fairly complicated URL
>> rewrite
>> logic in HttpNetworkTransaction.
>>
>> BUG=None
>> TEST=built and ran net_unittests
>>
>> Please review this at http://codereview.chromium.org/3041019/show
>>
>> Affected files:
>>  M chrome/browser/browser_main.cc
>>  M chrome/common/chrome_switches.h
>>  M chrome/common/chrome_switches.cc
>>  M net/http/http_network_session.h
>>  M net/http/http_network_session.cc
>>  M net/http/http_network_transaction.cc
>>  M net/spdy/spdy_network_transaction.cc
>>
>>
>>
>

[mbelshe, 2010-07-26 17:00:17]

On Mon, Jul 26, 2010 at 9:54 AM, Chris Bentzel <cbentzel@chromium.org>wrote:

> Would host-rules be sufficient for you?


98% of the time I use host-resolver-rules...

But there are a few times when I need to control the per-protocol ports,
which is where these come into play.

Mike



>
>
> On Mon, Jul 26, 2010 at 12:42 PM, Mike Belshe <mbelshe@google.com> wrote:
>
>> I do use these every now and then...
>>
>> Mike
>>
>>
>> On Mon, Jul 26, 2010 at 6:21 AM, <cbentzel@chromium.org> wrote:
>>
>>> Reviewers: Mike Belshe, eroman,
>>>
>>> Message:
>>> The only concern with dropping these two flags is that --host-rules does
>>> rules
>>> matching using the host, but never the port.
>>>
>>> If you are testing against a single server name, it doesn't seem to
>>> onerous to
>>> add the host name for matching to the new port. If multiple (possibly in
>>> the
>>> case of the SPDY in memory server?) I'll need to add host matches based
>>> on
>>> destination port before making this change.
>>>
>>> Description:
>>> Remove --testing-fixed-http-port and --testing-fixed-https-port.
>>>
>>> --host-rules seems like a reasonable enough substitute for these two
>>> command
>>> line flags, and removing them will simplifies the fairly complicated URL
>>> rewrite
>>> logic in HttpNetworkTransaction.
>>>
>>> BUG=None
>>> TEST=built and ran net_unittests
>>>
>>> Please review this at http://codereview.chromium.org/3041019/show
>>>
>>> Affected files:
>>>  M chrome/browser/browser_main.cc
>>>  M chrome/common/chrome_switches.h
>>>  M chrome/common/chrome_switches.cc
>>>  M net/http/http_network_session.h
>>>  M net/http/http_network_session.cc
>>>  M net/http/http_network_transaction.cc
>>>  M net/spdy/spdy_network_transaction.cc
>>>
>>>
>>>
>>
>

[cbentzel, 2010-07-26 17:01:49]

OK. I'll hold off on this then. Darn, was hoping to simplify URL rewrite
paths in HttpNetworkTransaction.

On Mon, Jul 26, 2010 at 1:00 PM, Mike Belshe <mbelshe@google.com> wrote:

> On Mon, Jul 26, 2010 at 9:54 AM, Chris Bentzel <cbentzel@chromium.org>wrote:
>
>> Would host-rules be sufficient for you?
>
>
> 98% of the time I use host-resolver-rules...
>
> But there are a few times when I need to control the per-protocol ports,
> which is where these come into play.
>
> Mike
>
>
>
>>
>>
>> On Mon, Jul 26, 2010 at 12:42 PM, Mike Belshe <mbelshe@google.com> wrote:
>>
>>> I do use these every now and then...
>>>
>>> Mike
>>>
>>>
>>> On Mon, Jul 26, 2010 at 6:21 AM, <cbentzel@chromium.org> wrote:
>>>
>>>> Reviewers: Mike Belshe, eroman,
>>>>
>>>> Message:
>>>> The only concern with dropping these two flags is that --host-rules does
>>>> rules
>>>> matching using the host, but never the port.
>>>>
>>>> If you are testing against a single server name, it doesn't seem to
>>>> onerous to
>>>> add the host name for matching to the new port. If multiple (possibly in
>>>> the
>>>> case of the SPDY in memory server?) I'll need to add host matches based
>>>> on
>>>> destination port before making this change.
>>>>
>>>> Description:
>>>> Remove --testing-fixed-http-port and --testing-fixed-https-port.
>>>>
>>>> --host-rules seems like a reasonable enough substitute for these two
>>>> command
>>>> line flags, and removing them will simplifies the fairly complicated URL
>>>> rewrite
>>>> logic in HttpNetworkTransaction.
>>>>
>>>> BUG=None
>>>> TEST=built and ran net_unittests
>>>>
>>>> Please review this at http://codereview.chromium.org/3041019/show
>>>>
>>>> Affected files:
>>>>  M chrome/browser/browser_main.cc
>>>>  M chrome/common/chrome_switches.h
>>>>  M chrome/common/chrome_switches.cc
>>>>  M net/http/http_network_session.h
>>>>  M net/http/http_network_session.cc
>>>>  M net/http/http_network_transaction.cc
>>>>  M net/spdy/spdy_network_transaction.cc
>>>>
>>>>
>>>>
>>>
>>
>

[eroman, 2010-07-26 17:02:07]

lgtm.

i'll let you and mike duke it out.

On Mon, Jul 26, 2010 at 10:00 AM, Mike Belshe <mbelshe@google.com> wrote:
> On Mon, Jul 26, 2010 at 9:54 AM, Chris Bentzel <cbentzel@chromium.org>
> wrote:
>>
>> Would host-rules be sufficient for you?
>
> 98% of the time I use host-resolver-rules...
> But there are a few times when I need to control the per-protocol ports,
> which is where these come into play.
> Mike
>
>>
>> On Mon, Jul 26, 2010 at 12:42 PM, Mike Belshe <mbelshe@google.com> wrote:
>>>
>>> I do use these every now and then...
>>> Mike
>>>
>>> On Mon, Jul 26, 2010 at 6:21 AM, <cbentzel@chromium.org> wrote:
>>>>
>>>> Reviewers: Mike Belshe, eroman,
>>>>
>>>> Message:
>>>> The only concern with dropping these two flags is that --host-rules does
>>>> rules
>>>> matching using the host, but never the port.
>>>>
>>>> If you are testing against a single server name, it doesn't seem to
>>>> onerous to
>>>> add the host name for matching to the new port. If multiple (possibly in
>>>> the
>>>> case of the SPDY in memory server?) I'll need to add host matches based
>>>> on
>>>> destination port before making this change.
>>>>
>>>> Description:
>>>> Remove --testing-fixed-http-port and --testing-fixed-https-port.
>>>>
>>>> --host-rules seems like a reasonable enough substitute for these two
>>>> command
>>>> line flags, and removing them will simplifies the fairly complicated URL
>>>> rewrite
>>>> logic in HttpNetworkTransaction.
>>>>
>>>> BUG=None
>>>> TEST=built and ran net_unittests
>>>>
>>>> Please review this at http://codereview.chromium.org/3041019/show
>>>>
>>>> Affected files:
>>>>  M chrome/browser/browser_main.cc
>>>>  M chrome/common/chrome_switches.h
>>>>  M chrome/common/chrome_switches.cc
>>>>  M net/http/http_network_session.h
>>>>  M net/http/http_network_session.cc
>>>>  M net/http/http_network_transaction.cc
>>>>  M net/spdy/spdy_network_transaction.cc
>>>>
>>>>
>>>
>>
>
>

[mbelshe, 2010-07-26 17:35:13]

LGTM; if I'm the only one that ever uses this, I can hack around it when I
need to.

Mike


On Mon, Jul 26, 2010 at 10:00 AM, Mike Belshe <mbelshe@google.com> wrote:

> On Mon, Jul 26, 2010 at 9:54 AM, Chris Bentzel <cbentzel@chromium.org>wrote:
>
>> Would host-rules be sufficient for you?
>
>
> 98% of the time I use host-resolver-rules...
>
> But there are a few times when I need to control the per-protocol ports,
> which is where these come into play.
>
> Mike
>
>
>
>>
>>
>> On Mon, Jul 26, 2010 at 12:42 PM, Mike Belshe <mbelshe@google.com> wrote:
>>
>>> I do use these every now and then...
>>>
>>> Mike
>>>
>>>
>>> On Mon, Jul 26, 2010 at 6:21 AM, <cbentzel@chromium.org> wrote:
>>>
>>>> Reviewers: Mike Belshe, eroman,
>>>>
>>>> Message:
>>>> The only concern with dropping these two flags is that --host-rules does
>>>> rules
>>>> matching using the host, but never the port.
>>>>
>>>> If you are testing against a single server name, it doesn't seem to
>>>> onerous to
>>>> add the host name for matching to the new port. If multiple (possibly in
>>>> the
>>>> case of the SPDY in memory server?) I'll need to add host matches based
>>>> on
>>>> destination port before making this change.
>>>>
>>>> Description:
>>>> Remove --testing-fixed-http-port and --testing-fixed-https-port.
>>>>
>>>> --host-rules seems like a reasonable enough substitute for these two
>>>> command
>>>> line flags, and removing them will simplifies the fairly complicated URL
>>>> rewrite
>>>> logic in HttpNetworkTransaction.
>>>>
>>>> BUG=None
>>>> TEST=built and ran net_unittests
>>>>
>>>> Please review this at http://codereview.chromium.org/3041019/show
>>>>
>>>> Affected files:
>>>>  M chrome/browser/browser_main.cc
>>>>  M chrome/common/chrome_switches.h
>>>>  M chrome/common/chrome_switches.cc
>>>>  M net/http/http_network_session.h
>>>>  M net/http/http_network_session.cc
>>>>  M net/http/http_network_transaction.cc
>>>>  M net/spdy/spdy_network_transaction.cc
>>>>
>>>>
>>>>
>>>
>>
>