> On May 18, 2015, 10:04 p.m., Joris Van Remoortere wrote:
> > 3rdparty/libprocess/3rdparty/stout/tests/flags_tests.cpp, line 511
> > <https://reviews.apache.org/r/34193/diff/2/?file=963014#file963014line511>
> >
> >     Since we're in an implementation file, we can `using 
> > std::ostringstream;` and then just use `ostringstream` here.
> 
> Marco Massenzio wrote:
>     do I have to?
>     I don't particularly like the `using` thing (although appreciate its 
> usefulness) and avoid using (ha, a pun!) it unless the type is used in 
> several places, in which case, the gain is significant.
> 
> Joris Van Remoortere wrote:
>     I'm not sure if we have an explicit rule here for this. I just know if 
> someone else comes to review this code, they will point out the same thing. 
> Let's just go with consistency.
> 
> Marco Massenzio wrote:
>     I'm not sure I follow the logic here, Joris - I'll go along with it, as 
> an extra 20 keystrokes or so are not something I want to make a big issue 
> about, but it seems warped; I also note here, that I'm adding the `using` 
> clause some several 100's lines above where it's used, and that goes against 
> the whole concept of readability.

BTW - just came across this:
```
    std::string pad(PAD + width - line.size(), ' ');
    line += pad;

    size_t pos1 = 0, pos2 = 0;
    // ...
    while (pos2 != std::string::npos) {  
```
in flags.hpp - so it would appear that using namespace-scoped variables if just 
fine (at least in Stout)?
IMO this only makes sense if you use the same type/variable several times 
and/or typing up the full namespace is tedious: `std::` is actually pretty low 
overhead.


- Marco


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/34193/#review84181
-----------------------------------------------------------


On May 19, 2015, 10:31 p.m., Marco Massenzio wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/34193/
> -----------------------------------------------------------
> 
> (Updated May 19, 2015, 10:31 p.m.)
> 
> 
> Review request for mesos, Benjamin Hindman and Joris Van Remoortere.
> 
> 
> Bugs: MESOS-2711
>     https://issues.apache.org/jira/browse/MESOS-2711
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> Jira: MESOS-2711
> 
> Every program that uses stout's `BaseFlags` ends up
> re-implementing the `printUsage()` function, and adding
> a `bool help` (and associated --help flag); this functionality
> has now been refactored in the base class and is available everywhere.
> 
> This change attempts to be backward-compatible, so it
> does not alter the behavior of the program when --help is
> invoked (by, eg, automatically printing usage and exiting)
> but leaves up to the caller to check for `flags.help` and then
> decide what action to take.
> 
> There is now a default behavior for the "leader" ("Usage: <prog name> 
> [options]")
> but the client API allows to modify that too.
> 
> Note - anywhere I found the use of the `--help` flag the behavior was the 
> same: 
> print usage and exit (see also https://reviews.apache.org/r/34195).
> 
> 
> Diffs
> -----
> 
>   3rdparty/libprocess/3rdparty/stout/include/stout/flags/flags.hpp 
> fb383b463a99924483634eebf22bf34de318f920 
>   3rdparty/libprocess/3rdparty/stout/tests/flags_tests.cpp 
> 00281195b53d2597bdb46e3fe6cd9d46a5e9b1f1 
> 
> Diff: https://reviews.apache.org/r/34193/diff/
> 
> 
> Testing
> -------
> 
> make check
> 
> **NOTE** this change by itself breaks the build, because the --help is 
> redefined everywhere (16 places, as of last count): CL 34195 fixes that and 
> makes all build/tests pass.
> 
> 
> Thanks,
> 
> Marco Massenzio
> 
>

Reply via email to