All SSL UI tests work with ephemeral ports.

All SSL UI tests work with ephemeral ports.

This also changes the testserver text replacement hooks to accept base64 encoded pairs, and to support multiple replacement texts.

These changes were needed for some of the "ssl/top_frame.html" based tests, which contained a frame with an HTML page referencing three URLs on three tdifferent test servers.

BUG=56814
TEST=browser_tests --gtest_filter="*SSLUI*"

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=65819

[akalin, 2010-11-11 05:20:54]

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc
File net/test/test_server.cc (right):

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc#new...
net/test/test_server.cc:281: DCHECK_EQ(0U, (text_to_replace.size() % 2));
if you take a vector of pairs, this won't be necessary

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc#new...
net/test/test_server.cc:282: if (text_to_replace.empty()) {
this check isn't necessary -- if you fall through it behaves the same way

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.h
File net/test/test_server.h (right):

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.h#newc...
net/test/test_server.h:131: const std::vector<std::string>& text_to_replace,
what about taking a vector of pairs of strings?

http://codereview.chromium.org/4664009/diff/10001/net/tools/testserver/testse...
File net/tools/testserver/testserver.py (right):

http://codereview.chromium.org/4664009/diff/10001/net/tools/testserver/testse...
net/tools/testserver/testserver.py:588: # TODO(cbentzel): Do all string
substitutions in one pass? There are
i don't think doing it in one pass is worth it; i wouldn't even bother having
this as a todo

http://codereview.chromium.org/4664009/diff/10001/net/tools/testserver/testse...
net/tools/testserver/testserver.py:592: if len(replace_text_args) != 2:
add warning here?

[Paweł Hajdan Jr., 2010-11-11 10:40:44]

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc
File net/test/test_server.cc (right):

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc#new...
net/test/test_server.cc:281: DCHECK_EQ(0U, (text_to_replace.size() % 2));
On 2010/11/11 05:20:54, akalin wrote:
> if you take a vector of pairs, this won't be necessary

Yeah, this should really take a vector of pairs.

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc#new...
net/test/test_server.cc:297: new_file_path += (i == 0) ? "?" : "&";
I'm not sure if I understand the logic that decides between "?" and "&". The
first query parameter should have "?", and all following should have "&".
However, does i == 0 mean this is the first query string parameter, and i != 0
mean we already had one query string parameter?

http://codereview.chromium.org/4664009/diff/10001/net/tools/testserver/testse...
File net/tools/testserver/testserver.py (right):

http://codereview.chromium.org/4664009/diff/10001/net/tools/testserver/testse...
net/tools/testserver/testserver.py:588: # TODO(cbentzel): Do all string
substitutions in one pass? There are
On 2010/11/11 05:20:54, akalin wrote:
> i don't think doing it in one pass is worth it; i wouldn't even bother having
> this as a todo

Yeah.

http://codereview.chromium.org/4664009/diff/10001/net/tools/testserver/testse...
net/tools/testserver/testserver.py:592: if len(replace_text_args) != 2:
On 2010/11/11 05:20:54, akalin wrote:
> add warning here?

Just make it explode with an exception, to make sure it's noticed.

[cbentzel, 2010-11-11 15:34:09]

Thanks for the speedy and useful reviews!

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc
File net/test/test_server.cc (right):

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc#new...
net/test/test_server.cc:281: DCHECK_EQ(0U, (text_to_replace.size() % 2));
That made things cleaner, thanks for suggestion.

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc#new...
net/test/test_server.cc:282: if (text_to_replace.empty()) {
On 2010/11/11 05:20:54, akalin wrote:
> this check isn't necessary -- if you fall through it behaves the same way

Done.

http://codereview.chromium.org/4664009/diff/10001/net/test/test_server.cc#new...
net/test/test_server.cc:297: new_file_path += (i == 0) ? "?" : "&";
On 2010/11/11 10:40:45, Paweł Hajdan Jr. wrote:
> I'm not sure if I understand the logic that decides between "?" and "&". The
> first query parameter should have "?", and all following should have "&".
> However, does i == 0 mean this is the first query string parameter, and i != 0
> mean we already had one query string parameter?

That's correct. The new implementation should make it more clear.

http://codereview.chromium.org/4664009/diff/10001/net/tools/testserver/testse...
File net/tools/testserver/testserver.py (right):

http://codereview.chromium.org/4664009/diff/10001/net/tools/testserver/testse...
net/tools/testserver/testserver.py:588: # TODO(cbentzel): Do all string
substitutions in one pass? There are
On 2010/11/11 10:40:45, Paweł Hajdan Jr. wrote:
> On 2010/11/11 05:20:54, akalin wrote:
> > i don't think doing it in one pass is worth it; i wouldn't even bother
having
> > this as a todo
> 
> Yeah.

Done.

http://codereview.chromium.org/4664009/diff/10001/net/tools/testserver/testse...
net/tools/testserver/testserver.py:592: if len(replace_text_args) != 2:
On 2010/11/11 10:40:45, Paweł Hajdan Jr. wrote:
> On 2010/11/11 05:20:54, akalin wrote:
> > add warning here?
> 
> Just make it explode with an exception, to make sure it's noticed.

Done.

[Paweł Hajdan Jr., 2010-11-11 18:09:52]

LGTM

[akalin, 2010-11-11 19:35:40]

LGTM

On 2010/11/11 18:09:52, Paweł Hajdan Jr. wrote:
> LGTM