> On Dec. 14, 2017, 9:40 p.m., Benjamin Mahler wrote:
> > Can you link to some prior art? What do the RFCs say recommend, if 
> > anything? What do other http frameworks do?

According to [RFC3986](https://tools.ietf.org/html/rfc3986#section-3.3), every 
'/' is meaningful, i.e., `/path` and `/path/` have different number of 
segments: ["path"] vs. ["path", ""]. However in practice it is recommended to 
treat paths with leading and trailing slashes in the same way to avoid user 
confusion, see [google's 
recommendation](https://webmasters.googleblog.com/2010/04/to-slash-or-not-to-slash.html).
 It is also recommended to redirect one request to another to avoid duplicate 
content, but I don't think it is relevant for us.


- Alexander


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/64574/#review193848
-----------------------------------------------------------


On Dec. 13, 2017, 1:39 p.m., Alexander Rukletsov wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/64574/
> -----------------------------------------------------------
> 
> (Updated Dec. 13, 2017, 1:39 p.m.)
> 
> 
> Review request for mesos, Benjamin Bannier and Benjamin Mahler.
> 
> 
> Bugs: MESOS-5333
>     https://issues.apache.org/jira/browse/MESOS-5333
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Prior to this patch, adding a trailing '/' to a valid URL path, e.g.,
> "/state/", yielded a 404 response. This patch ensures that two URLs
> which differ only in trailing '/' produce the same result.
> 
> 
> Diffs
> -----
> 
>   3rdparty/libprocess/src/process.cpp 
> 75cf1d3b6d3d257ba9bc81c68017a74a6511cebf 
>   3rdparty/libprocess/src/tests/http_tests.cpp 
> 9daac715f0242921b7f9f5c20b3eb27f1be802d4 
> 
> 
> Diff: https://reviews.apache.org/r/64574/diff/1/
> 
> 
> Testing
> -------
> 
> Ensured the modified test fails without the fix.
> `make check` on Mac OS 10.11.6 and several Linux distributions.
> 
> 
> Thanks,
> 
> Alexander Rukletsov
> 
>

Reply via email to