Subject: Re: [Boost-build] Python port development
From: Gevorg Voskanyan (v_gevorg_at_[hidden])
Date: 2010-07-09 16:34:53
> It is not tricky when you know the solution :-).
> But as you can see neither your nor Gevorg, did this right.
Yep, didn't consider cross-compile at all :(
Thanks for pointing out that problem, in light of that the capture-target-output
I wrote might not be that much useful...
> BTW: why do you need there *2? In order to get negative value as some
> compilers accept 0 as valid array size.
> Also it is simple to check one specific case sizeof >= KnownValue,
> but lets say you want to know that is the size of some structure,
> with unknown expected size.
> So you need to write a loop that compiles this for several cases and then run
> increasing each time
> twice and then run binary search... Not so simple at all.
So much pain for what? Defining some macro which carries information already
available and queryable at compile time? Why would a C++ library ever need the
sizeof of some type to do conditional compilation on it? Why would it need
either SIZE_T_IS_UINT or SIZE_T_IS_ULONG to be defined as appropriate? Oh yes,
templates weren't implemented that good or at all in the compilers when the
library began to be written. And then you have C libraries... But anyway it is
very disturbing having to deal with such things now. But that is the way things
are, and need to be supported, sigh.
Sorry for this, I just expressed my fresh feelings from the experience of
previous couple of days.
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk