Re: [cpp-netlib-devel] Compile-time Reduction Effort

[Dean M. B. <mik...@gm...>]

On Tue, Oct 26, 2010 at 2:59 AM, Glyn Matthews <gly...@gm...> wrote:
> Hi,
>
> On 25 October 2010 16:32, Nelson, Erik - 2 <eri...@ba...>
> wrote:
>>
>> > Dean Michael Berris wrote on Monday, October 25, 2010
>>
>> > Now there are still a lot of other things I can do to start reducing
>> > compile-times. One is incorporating precompiled headers into the
>> > build, and start pre-compiling almost every single header in the
>> > library.
>>
>> Pre-compiled headers seem like they are going to be a mess, and
>> conceptually similar to going to a non-header-only implementation.  I
>> like the idea of header-only, but maybe a more library-based one would
>> help the compile time.
>
> I agree with this point.  IMO, although I agree that the header-only
> approach has merits, it seems that since we're already requiring that users
> link against Boost's system, regex and date_time libraries, we don't seem to
> be gaining much from this approach anyway.

Actually, I'll work on removing the dependency on Boost.Regex soon and
use Boost.Xpressive instead to remove the link-time requirement to it.
Date_time also has an option to inline everything (IIRC) while
Boost.System will get a macro option to "inline" it.

So pretty soon cpp-netlib can really be header-only depending on Boost.

> Including the compile time
> problem and making cpp-netlib strictly being header-only has very few
> benefits.  I would suggest a better use of our focus should be on
> portability.

Yeah, the compile time problem is really brought in by Boost.Spirit's
Qi (and consequently Boost.Fusion and Boost.Proto). Although it is
portable already as it is now, right?

> Additionally, I have been doing a lot of development in embedded systems,
> where the long compile times and large binaries that are generated pretty
> much rules out using cpp-netlib.  This is a shame, because this is an area
> where cpp-netlib can provide a lot of benefits to developers.
>

Are you not stripping the binaries created by cpp-netlib? In Debug
mode, yes you will have problems with the debugging symbols eating up
the output binary.

Also, since we still depend on Boost.Lexical_cast that does depend on
the STL io streams library, that will cause the binaries to be pretty
large. We might want to try using Jeroen's construe_cast in places
where we use lexical_cast.

The long compile times can really be addressed by precompiled headers
especially if you're using GCC. Clang, well not so much because
precompiled headers support is still a little ... well... iffy at this
time. :D

>>
>> > Has anyone played around with CMake and how to make precompiled
>> > headers part of the build process? I have seen some Google results on
>> > the matter but I currently have no time/patience to muck around with
>> > that.
>> >
>>
>> Nothing else in Boost requires CMake, as far as I know.  Boost-devel has
>> had some rumblings about a change, but right now Boost.Build is the
>> Boost standard.
>
> Sadly ;)
>

Yeah, it is sad. :D

>>
>> > I'm nearing the completion of the asynchronous server that invokes
>> > request handlers in a thread pool and although the compile times have
>> > been reduced in the HTTP client side of things I would really
>> > appreciate some CMake kung-fu from people on the list to incorporate
>> > precompiled-headers (so that we can make that part of the install step
>> > in 0.8).
>> >
>>
>> Compile time *is* an important issue, but the beauty of a header library
>> is that there is no install step- you just copy the files into your
>> directory tree.
>
> But there is an install step, because of the dependencies that cpp-netlib
> has - and will have: in developing the XMPP client I'll be linking to an XML
> parser library and TLS library.  Other protocol implementations will bring
> other dependencies and we'll have to deal with those too.

Yeah, this is going to be the test of having an install step. :D

> Regards,

Thanks for the comments Glyn, hope mine help convince you that
precompiled headers and changes to cpp-netlib will be good for users
(and embedded systems). :)

-- 
Dean Michael Berris
deanberris.com