> On Sept. 8, 2015, 3:51 p.m., Artem Harutyunyan wrote:
> > 3rdparty/libprocess/3rdparty/stout/include/stout/windows.hpp, lines 152-159
> > <https://reviews.apache.org/r/37032/diff/3/?file=1064256#file1064256line152>
> >
> >     Do we really need to leave this up to the user to decide? What's the 
> > benefit of one versus the other? Can't we just stick to one option and just 
> > implement that one? 
> >     
> >     If not, is there a build time configuration flag that can be used to 
> > change this?
> 
> Alex Clemmer wrote:
>     I provided this (1) because I could see use cases for both permission 
> semantics (both for us and for downstream clients _e.g._ DCOS clients who are 
> concerned about security or something), (2) because making this configurable 
> at runtime would likely mean augmenting or refactoring the existing API for 
> things like `chown`, and (3) because a "final" solution is probably out of 
> scope for the MVP.
>     
>     Happy to change this, but for now I think we should probably ship it and 
> re-evaluate later.
>     
>     Also, to answer your question about specifying this at compile time, yes, 
> you can just pass a flag like `-DSTRICT_OTHER_PERMISSIONS` and it should 
> "just work." This isn't wired up in CMake but that would be trivial to do.

I suggest for now to get rid of the flags and opt for strict-ness (no fallback) 
by default.  Being strict means places in the codebase that try to use 
"group/other" permissions will be more likely to error out (i.e. access 
denied).  We can then catch those cases individually and deal with them.

We should also create a JIRA to track this, in-case anything breaks as a result 
(on Windows).


- Joseph


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/37032/#review98092
-----------------------------------------------------------


On Sept. 10, 2015, 12:36 p.m., Alex Clemmer wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/37032/
> -----------------------------------------------------------
> 
> (Updated Sept. 10, 2015, 12:36 p.m.)
> 
> 
> Review request for mesos, Benjamin Hindman, Artem Harutyunyan, Joris Van 
> Remoortere, and Joseph Wu.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Extend permissions.hpp to work on both Windows and POSIX.
> 
> 
> Diffs
> -----
> 
>   3rdparty/libprocess/3rdparty/stout/include/Makefile.am 
> 8853f92fcfcff81d0a3197bade02110685fa0325 
>   3rdparty/libprocess/3rdparty/stout/include/stout/os/permissions.hpp 
> 196c3f5fac7c3526924f2bea03c06d1fbce25c61 
>   3rdparty/libprocess/3rdparty/stout/include/stout/os/posix/permissions.hpp 
> 98f0b3c8e55190df87d6a581667e21b31ac044bc 
>   3rdparty/libprocess/3rdparty/stout/include/stout/os/windows/permissions.hpp 
> daed4b4e9c76d6e7c043a1fa3a46ab1f3db95f48 
>   3rdparty/libprocess/3rdparty/stout/include/stout/windows.hpp 
> 7ab75ece44ab4b0cc42f992daf1101d0faf80b1f 
> 
> Diff: https://reviews.apache.org/r/37032/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Alex Clemmer
> 
>

Reply via email to