Remove redundant mapping of net errors to strings.

Remove redundant mapping of net errors to strings.

The new list does not have the leading "net::ERR_", to slightly reduce
the size of net when compiled.  This reduces the size of libcronet.so
by about 8k, with symbols stripped.

Also add a function to get network error strings without the leading,
"net::" as a number of consumers were removing it with duplicated code.

This CL slight breaks NetLog loading functionality - loading old logs
in new versions  will result in error codes missing the leading "ERR_",
and the other direction results in error codes having an extra leading
"ERR_".

BUG=399025
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=288065

[mmenke, 2014-08-01 15:59:53]

I slightly changed the format of NetLog dumps, increasing them all by about 1k. 
That's not strictly necessary, but passing around 3 different string
representations for each network error code just seems a bit excessive.  If you
disagree, we can clip off the "ERR_" again (Adding a third function for that
format is a bit too much, methinks).

If we go with the change, I'll announce it on net-dev, and maybe chromium-dev as
well.  It only really matters when using filters, and dumps from different
versions of Chrome.

[eroman, 2014-08-05 22:54:51]

Don't forget to update this too:

https://code.google.com/p/chromium/codesearch#chromium/src/chrome/browser/res...

The only breakage that happens when loading old logs is that you need to know to
search for the corresponding new names right? The rest I imagine just works
since it uses the mappings.

LGTM

https://codereview.chromium.org/432553003/diff/120001/chrome/common/localized...
File chrome/common/localized_error.cc (left):

https://codereview.chromium.org/432553003/diff/120001/chrome/common/localized...
chrome/common/localized_error.cc:584: ascii_error_string.erase(0, 5);
wow, this was duplicated in lots of places!

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.cc
File net/base/net_errors.cc (right):

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.cc#...
net/base/net_errors.cc:40: return "<unknown>";
Should this be prefixed with ERR_ too?

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.cc#...
net/base/net_errors.cc:42: return std::string("ERR_").append(error_string);
nit: below you use operator+ to concatenate strings, yet here append(). For
consistency I suggest sticking with one style (operator+).

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.cc#...
net/base/net_errors.cc:54: return (error <= ERR_CERT_BEGIN && error >
ERR_CERT_END) ||
I presume this is unchanged, did not review.

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.h
File net/base/net_errors.h (right):

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.h#n...
net/base/net_errors.h:33: // Same as above, but leaves off the leading "net::".
"Same as below", or move this down ;)

[mmenke, 2014-08-06 16:40:56]

For external consumers, this CL basically makes two changes:  It adds a function
"ErrorToShortString", which returns the network error code without the leading
"net::", since a lot of consumers were doing that themselves, and it makes
ErrorToString return a string instead of a char*.  Changes elsewhere are largely
mechanical, to account for these two changes.

[+thestig]:  Please review chrome/browser/errorpage_browsertest.cc and
chrome/common/localized_error.cc.
[+fsamuel]:  Please review chrome/browser/guest_view/web_view/
[+jam]:  Please review content/child
[+armansito]:  Please review chromeos/network/onc/ and device/bluetooth.

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.cc
File net/base/net_errors.cc (right):

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.cc#...
net/base/net_errors.cc:40: return "<unknown>";
On 2014/08/05 22:54:51, eroman wrote:
> Should this be prefixed with ERR_ too?

Done.  I did that initially, but it looks kinda weird.  But I guess it should
look kinda weird.

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.cc#...
net/base/net_errors.cc:42: return std::string("ERR_").append(error_string);
On 2014/08/05 22:54:50, eroman wrote:
> nit: below you use operator+ to concatenate strings, yet here append(). For
> consistency I suggest sticking with one style (operator+).

Done.

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.cc#...
net/base/net_errors.cc:54: return (error <= ERR_CERT_BEGIN && error >
ERR_CERT_END) ||
On 2014/08/05 22:54:50, eroman wrote:
> I presume this is unchanged, did not review.

Correct.  I just moved it, since it was incorrectly inlined in the header.

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.h
File net/base/net_errors.h (right):

https://codereview.chromium.org/432553003/diff/120001/net/base/net_errors.h#n...
net/base/net_errors.h:33: // Same as above, but leaves off the leading "net::".
On 2014/08/05 22:54:51, eroman wrote:
> "Same as below", or move this down ;)

Moved it down (And its definition as well), "same as below" just seems weird.

[Fady Samuel, 2014-08-06 18:39:47]

guest_view lgtm

[Lei Zhang, 2014-08-06 18:39:52]

*.cc files in chrome/ lgtm

[armansito, 2014-08-06 20:46:15]

device/bluetooth & chromeos/network lgtm

[jam, 2014-08-07 05:48:45]

lgtm

[mmenke, 2014-08-07 11:28:58]

The CQ bit was checked by mmenke@chromium.org

[commit-bot: I haz the power, 2014-08-07 11:30:00]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/mmenke@chromium.org/432553003/180001

[commit-bot: I haz the power, 2014-08-07 16:12:13]

Change committed as 288065