RESOLVED WONTFIX 87002
SecurityOrigin returns 0 as port when it is the default port for the protocol
https://bugs.webkit.org/show_bug.cgi?id=87002
Summary SecurityOrigin returns 0 as port when it is the default port for the protocol
Thiago Marcos P. Santos
Reported 2012-05-21 04:28:33 PDT
It should return the correct value instead, like 80 for http, 443 for https, etc.
Attachments
Patch (4.42 KB, patch)
2012-05-21 06:03 PDT, Thiago Marcos P. Santos
no flags
Patch (5.21 KB, patch)
2012-05-21 07:41 PDT, Thiago Marcos P. Santos
no flags
Patch (6.36 KB, patch)
2012-05-21 09:24 PDT, Thiago Marcos P. Santos
no flags
Thiago Marcos P. Santos
Comment 1 2012-05-21 06:03:10 PDT
Build Bot
Comment 2 2012-05-21 06:48:59 PDT
Thiago Marcos P. Santos
Comment 3 2012-05-21 07:41:32 PDT
Build Bot
Comment 4 2012-05-21 08:18:18 PDT
Adam Barth
Comment 5 2012-05-21 09:15:49 PDT
Comment on attachment 143032 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=143032&action=review Please add an API test that shows the change in behavior. These tests are easy to write for Chromium, but you should feel free to write one for a different port if you prefer. Would you be willing to list in this bug which APIs are changing in which port? There's some risk of a compatibility problem. For example, if some client of the API is trying to construct storage identifiers manually, they might get a different answer after this patch... > Source/WebCore/WebCore.order:6589 > +__ZNK7WebCore14SecurityOrigin4portEv You probably need to add this line to WebCore.exp.in to fix the apple-mac build. > Source/WebCore/page/SecurityOrigin.h:68 > - unsigned short port() const { return m_port; } > + unsigned short port() const; I wonder if we should change the name of this function so it doesn't look like an accessor for m_port. Perhaps we should call it portNumber()?
Adam Barth
Comment 6 2012-05-21 09:17:22 PDT
It's still slightly unclear to me whether this patch is solving a real problem. We need to balance the risk of changing an API against the benefit of doing so. I don't mind changing this in the Chromium API, but other folks might be more conservative about how they'd like their APIs to evolve.
Thiago Marcos P. Santos
Comment 7 2012-05-21 09:24:20 PDT
Thiago Marcos P. Santos
Comment 8 2012-05-21 09:47:34 PDT
It is too risky to change this method. Should be fixed on the API layer if needed.
Note You need to log in before you can comment on or make changes to this bug.