[webkit-dev] Super-slow websocket tests

[Fumitoshi Ukai (鵜飼文敏)]

On Thu, Jan 14, 2010 at 12:29 PM, Alexey Proskuryakov <ap at webkit.org> wrote:

>
> 13.01.2010, в 19:16, Fumitoshi Ukai (鵜飼文敏) написал(а):
>
> 6.87 secs: websocket/tests/frame-lengths.html
>>>
>>
>> This test verifies that packet size computation logic is correct for all
>> message sizes from 0 to 1025. We've had bugs in this area before, and it's
>> hard to ensure adequate coverage without testing all message sizes.
>>
>
> Sorry, I don't remember what bugs in this area.
>
>
> https://bugs.webkit.org/show_bug.cgi?id=30656
> https://bugs.webkit.org/show_bug.cgi?id=32203
>
>
These bugs are length-based framing, which isn't used in normal websocket
messaging yet. (we don't send these frames, and we should skip these
frames).
frame-lengths.html tries to send any length of text, but these messages are
delivered with \xFF-terminated frame (frame-type 0x00), not with
length-based frame (frame-type >0x80).
So, I don't think frame-lengths.html would catch bugs something like above
bugs...

-- 
ukai

>  Can we do test only boundary case?
>
>
> Perhaps. I'm not sure what case to consider being boundary.
>
> - WBR, Alexey Proskuryakov
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-dev/attachments/20100114/a696aca4/attachment.html>