> On April 28, 2015, 2:08 a.m., Ben Mahler wrote: > > docs/mesos-c++-style-guide.md, line 162 > > <https://reviews.apache.org/r/33558/diff/3/?file=942044#file942044line162> > > > > When would we use capture by reference? Seems prone to mistakes?
> When would we use capture by reference? We would use it when we know the lambda will be invoked within the scope. For example, if we're calling an algorithm (e.g. `std::sort`) that takes a lambda, we know that it'll be invoked within the scope. > Seems prone to mistakes? It is indeed prone to mistakes which is the motivation behind `*never* use default capture by reference`. > On April 28, 2015, 2:08 a.m., Ben Mahler wrote: > > docs/mesos-c++-style-guide.md, lines 206-208 > > <https://reviews.apache.org/r/33558/diff/3/?file=942044#file942044line206> > > > > What does it mean to pass a lambda to `socket.send`? Doesn't look like > > this is functionality provided on the existing socket we have, and I'm not > > sure what this would do.. provide a functor of the data? As this review is about style, I myself didn't dig into the validity of this example all that much. But perhaps we should find a more realistic example. > On April 28, 2015, 2:08 a.m., Ben Mahler wrote: > > docs/mesos-c++-style-guide.md, lines 237-294 > > <https://reviews.apache.org/r/33558/diff/3/?file=942044#file942044line237> > > > > Hm.. this look a bit hard to read to me, you might think that the > > capture list is an argument as you scan case 1, for example, and it might > > be hard to distinguish the lambda from other arguments (consider if > > socket.send takes more arguments). > > > > Would it be better to illustrate the wrapping via a variable? > > > > ``` > > auto lambda = [&capture1, &capture2, &capture3] > > (const T1& p1, const T2& p2, const T3& p3) { > > ...; > > }; > > ``` > > > > One last note, it seeems nice in many cases to keep the opening > > parenthesis with the parameters rather than against the end of the capture > > list (e.g. as I did above). I realize that's not clean to do when you have > > to wrap the parameters as well though, so maybe not :) +1 to illustrating the examples via a variable. > One last note, it seeems nice in many cases to keep the opening parenthesis > with the parameters rather than against the end of the capture list (e.g. as > I did above). I realize that's not clean to do when you have to wrap the > parameters as well though, so maybe not :) I don't like that it would deviate further from how we format parameters of regular functions. In specific, we don't format functions like this: ```cpp SomeLongReturnType SomeLongFunctionName (const T1& p1, const T2& p2, const T3& p3) { ... } ``` we format it like this: ```cpp SomeLongReturnType SomeLongFunctionName( const T1& p1, const T2& p2, const T3& p3) { ... } ``` By "further" above I'm referring to the fact that we put a newline between `) {` for regular functions whereas for lambdas we don't. - Michael ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/33558/#review81769 ----------------------------------------------------------- On April 26, 2015, 8:26 p.m., Benjamin Hindman wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/33558/ > ----------------------------------------------------------- > > (Updated April 26, 2015, 8:26 p.m.) > > > Review request for mesos, Ben Mahler, Joris Van Remoortere, Michael Park, and > Vinod Kone. > > > Repository: mesos > > > Description > ------- > > See summary. Note that there will be a follow up review which enacts this > style in the code base. > > > Diffs > ----- > > docs/mesos-c++-style-guide.md fe98f90ad0b0f5dd38af97e85062e90cee8de99e > > Diff: https://reviews.apache.org/r/33558/diff/ > > > Testing > ------- > > N/A > > > Thanks, > > Benjamin Hindman > >
