Boost logo

Geometry :

Subject: [ggl] Tests compilation broken (temporarily)
From: Mateusz Loskot (mateusz)
Date: 2009-04-30 05:14:53

Barend Gehrels wrote:
> Hi Mateusz,
>>> OK, everything is back to compilable state, tests and examples.
>> Perfect! Will try it soon.
> I scanned now for your changes and there are a lot! Thanks for all of them!

My pleasure, my fun :-)

> I like your additions for styles and the MS property sheets including
> Boost. I encountered those sheets recently, using soci (nice library),
> and wanted to add them, so you added them now, nice!

Yes, I use Property Sheets quite often and added them to SOCI.
However, there are number of drawbacks. Simply, they are static
and mostly users have different environments, so Linux ./configure-like
dynamic configurator is required.

Recently, I've been thinking of adding CMake configuration for GGL,
However, it would make sense to have CMake only if we used custom
CMake framework/macros added recently to Boost (looks like Boost is
migrating to CMake?)
Otherwise, we'd need to craft CMake config manually and get it lost
after we move GGL to Boost.

So, I'd consider the property sheets as a temporary solution.
It works for now and we don't need user-friendly building configuration,
I think.

Later, I'd vote for CMake - multiplatform meta-builder.

> Maybe we can add
> the two dependant libraries in the samples (GD and shapelib) to them or
> to a special version of them. That would be very handy.

You're reading my mind :-)
Good idea, especially they are a few files only.
I can do that during the weekend.

> I also like your careful exploration and corrections of header files
> like size_t -> std::size_t, etc, I was not always too keen on that.

OK. I like strict compilation mode, so I fix such small issues
if you don't mind.

By the way, have you tried to compile GGL with Comeau C/C++ ?

> One other thing (thought I mentioned it earlier), the
> intersection_polygon will be revised completely,

Yes I remember, but even that I've micro-revised it by the way of

> sorry about your changes in that file.

No problem at all!

> It will use the new segment intersection routine
> and then a sort of unified intersection/union approach...


Best regards,

Mateusz Loskot,

Geometry list run by mateusz at