Re: [lmi] Clang fixes

[Vadim Zeitlin]

On Sat, 26 Mar 2016 14:30:53 +0000 Greg Chicares <address@hidden> wrote:

GC> I'll look at this later. BTW, above you mention:
GC>   https://github.com/vadz/lmi/pull/14
GC> and then later you wrote:
GC> > 
https://github.com/vadz/lmi/pull/9/commits/a77ace84929150b3694e3ad2be9d3b99777a0fc8
GC> Which set(s) of patches should I examine?

 Well, ideally I'd like you to examine all pull requests, i.e. all items at
https://github.com/vadz/lmi/pulls but, to help you with prioritizing them,
I've put "high priority" tag on some of them as these are the ones that
prevent me from cross-compiling lmi currently. As you can see, I've also
put "simple" tag on the PRs which are, IMHO, simple to apply, i.e.
shouldn't require much time to review. Of course, this just expresses my
own estimations and I could be wrong...

GC> 9 and then 14, in numerical order?

 No, the numerical order definitely doesn't matter. So far the PRs are all
independent and can be applied in any order, I'll mention in the PR
comments if there is a dependency on another PR if this is ever needed. And
hopefully the tags mentioned above can help you to prioritize them a
little.


[getopt discussion]
GC> Here we only seem to disagree, but don't really. You're addressing getopt's 
API,
GC> while I'm addressing its CLI.

 Sorry, I've completely misunderstood you (I knew I must have been missing
something but I just couldn't understand what it was...). Of course the
user visible command line syntax must be the same as understood by getopt,
this is a de facto standard since a long time and nothing else makes sense
(except, possibly, minor deviations such as supporting "/" in addition to
"-" for the short options under MSW where this is very common). In any
case, I never even thought about proposing to change the current lmi
CLI which will, of course, remain the same.

 I only dislike getopt API. Docopt, that I mentioned before, takes a
completely different "declarative" approach which a lot of people like. But
personally I'd use something simpler just as gladly. As I said, there is
unfortunately nothing that is really perfect right now, but there is one
simple, header-only C++11 library which is close to what I think would be
ideal: https://github.com/philsquared/Clara but, as you can see, there is
no documentation nor even examples which is one of the problems with it.
If you'd like to discuss this further, I'd be glad to do it, but I just
don't want to waste your time with this right now if you don't plan to do
anything getopt in the immediate future anyhow, I'd rather deal with the
current patches first before thinking about making another one. But, just
not to forget about it, I've added a note about this to
https://github.com/vadz/lmi/issues

 And sorry again for misunderstanding you in the previous message,
VZ