Allow multiple identical Location and Content-Disposition headers

Allow broken servers to send us multiple identical Location and
Content-Disposition headers.

R=willchan@chromium.org
BUG=103618
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=128041

[mmenke, 2012-03-20 17:37:27]

The mac_rel compile failures look to be happening on all tryjobs, not just mine,
so not a problem with my CL.


Still want to do something when there's a comma in Content-Disposition headers
in a followup CL.  I think truncating everything after the comma is the simplest
and most robust approach.

If we have a comma in a file name or date, just cuts off the rest of the field,
which should be fine.  If we have two entire content-disposition fields, just
takes the first one.

Ideally, we could allow commas in dates (And maybe file names), while truncating
at the comma in other cases, either taking the first or the second half of the
field, but I don't think that's worth the effort.

[mmenke, 2012-03-20 18:05:27]

On 2012/03/20 17:37:27, Matt Menke wrote:
> The mac_rel compile failures look to be happening on all tryjobs, not just
mine,
> so not a problem with my CL.
> 
> 
> Still want to do something when there's a comma in Content-Disposition headers
> in a followup CL.  I think truncating everything after the comma is the
simplest
> and most robust approach.
> 
> If we have a comma in a file name or date, just cuts off the rest of the
field,
> which should be fine.  If we have two entire content-disposition fields, just
> takes the first one.
> 
> Ideally, we could allow commas in dates (And maybe file names), while
truncating
> at the comma in other cases, either taking the first or the second half of the
> field, but I don't think that's worth the effort.

Oh, and a bit off topic, but I'm thinking I'll move all the
HTTPNetworkTransaction tests that have nothing to do with SPDY into a single new
file so we don't have 3 copies of them all, unless there's a real need for them.

[willchan no longer on Chromium, 2012-03-20 18:09:00]

I agree with your assessment, lgtm. I'm adding Julian in case he wants to
comment.

[willchan no longer on Chromium, 2012-03-20 18:10:33]

Matt, please talk with rch/rtenneti before you do this. I assume you'll do
that in a separate changelist if you do it. And if you do, it's probably
easier to move the SPDY tests out of the
http_network_transaction_unittest.cc file rather than the vice versa.

On Tue, Mar 20, 2012 at 11:05 AM, <mmenke@chromium.org> wrote:

> On 2012/03/20 17:37:27, Matt Menke wrote:
>
>> The mac_rel compile failures look to be happening on all tryjobs, not just
>>
> mine,
>
>> so not a problem with my CL.
>>
>
>
>  Still want to do something when there's a comma in Content-Disposition
>> headers
>> in a followup CL.  I think truncating everything after the comma is the
>>
> simplest
>
>> and most robust approach.
>>
>
>  If we have a comma in a file name or date, just cuts off the rest of the
>>
> field,
>
>> which should be fine.  If we have two entire content-disposition fields,
>> just
>> takes the first one.
>>
>
>  Ideally, we could allow commas in dates (And maybe file names), while
>>
> truncating
>
>> at the comma in other cases, either taking the first or the second half
>> of the
>> field, but I don't think that's worth the effort.
>>
>
> Oh, and a bit off topic, but I'm thinking I'll move all the
> HTTPNetworkTransaction tests that have nothing to do with SPDY into a
> single new
> file so we don't have 3 copies of them all, unless there's a real need for
> them.
>
>
http://codereview.chromium.**org/9757002/<http://codereview.chromium.org/9757...
>

[mmenke1, 2012-03-20 18:11:46]

On 2012/03/20 18:10:33, willchan wrote:
> Matt, please talk with rch/rtenneti before you do this. I assume you'll do
> that in a separate changelist if you do it. And if you do, it's probably
> easier to move the SPDY tests out of the
> http_network_transaction_unittest.cc file rather than the vice versa.
> 
> On Tue, Mar 20, 2012 at 11:05 AM, <mailto:mmenke@chromium.org> wrote:
> 
> > On 2012/03/20 17:37:27, Matt Menke wrote:
> >
> >> The mac_rel compile failures look to be happening on all tryjobs, not just
> >>
> > mine,
> >
> >> so not a problem with my CL.
> >>
> >
> >
> >  Still want to do something when there's a comma in Content-Disposition
> >> headers
> >> in a followup CL.  I think truncating everything after the comma is the
> >>
> > simplest
> >
> >> and most robust approach.
> >>
> >
> >  If we have a comma in a file name or date, just cuts off the rest of the
> >>
> > field,
> >
> >> which should be fine.  If we have two entire content-disposition fields,
> >> just
> >> takes the first one.
> >>
> >
> >  Ideally, we could allow commas in dates (And maybe file names), while
> >>
> > truncating
> >
> >> at the comma in other cases, either taking the first or the second half
> >> of the
> >> field, but I don't think that's worth the effort.
> >>
> >
> > Oh, and a bit off topic, but I'm thinking I'll move all the
> > HTTPNetworkTransaction tests that have nothing to do with SPDY into a
> > single new
> > file so we don't have 3 copies of them all, unless there's a real need for
> > them.
> >
> >
>
http://codereview.chromium.**org/9757002/%3Chttp://codereview.chromium.org/97...>
> >

Yea, completely separate CL.  Was just staring at those 3 nearly identical 340k
files and scratching my head as I modified them all.

[mmenke, 2012-03-21 12:57:22]

I'm planning to land this this afternoon, and then submit followup CL for
review.

[commit-bot: I haz the power, 2012-03-21 17:52:27]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/mmenke@chromium.org/9757002/11004

[commit-bot: I haz the power, 2012-03-21 20:11:03]

Change committed as 128041

[mmenke, 2012-03-23 12:07:12]

[+julian]

Looks like the wrong email address for you was added to this bug.