Re: [Plplot-devel] Current status of our CBS

[Arjen M. <arj...@wl...>]

Alan W. Irwin wrote:

>Here is the current status of our CBS.
>
>Our CBS is virtually complete.  Here are the only exceptions I am aware of.
>Only the first of them is relevant on Linux.
>
>* octave is currently missing from our CBS.  That component is similar to
>Tcl/Tk with lots of straightforward details so I expect it will take a while
>to implement, but I don't expect any serious show-stoppers.  Andrew Ross has
>kindly volunteered to do the job so he will have the honor of finishing off
>our CBS on Linux.
>
>* The aqt device driver for Mac OS X is a work in progress, but it sounds
>like Hazen is close.
>
>* We need a windows-savvy volunteer to implement the
>sys/win32/msdev/src/win3.cpp device driver once bare windows works.
>
>* if CMake works for the djgpp platform (a big if), then Andrew Roach will
>undoubtedly want to implement the sys/dos/djgpp/src/gnusvga.c device driver.
>
>The various non-Linux platforms are beginning to round into shape thanks
>to the work of others.
>
>* On bare windows (visual c++ 2005), Werner tells me that at least the
>PLplot core is close to building. The one showstopper was how the math
>library was treated, but I think I have now fixed that so I hope it will
>build on the next iteration for him. Once that milestone occurs, there will
>probably be some considerable extra effort required to expand to other
>language interfaces (e.g., python and java), get examples working, etc., but
>the bare windows platform is looking quite promising.
>
>  
>
As Andy (aendey@..) reported, the PLplot.dll that gets built for bare 
windows seems
to be missing a few wrappers. This means that some examples do not compile.
I want to look into that this weekend.

And I am having trouble with CMake on Windows, I will report that on the
CMake list.

>* PLplot has built on Cygwin and MinGW in the past for Werner, but I have
>forgotten the completeness of the PLplot language interfaces he was testing,
>and how far he tested the examples.
>
>* As we learned recently from Hazen, PLplot builds okay on Mac except for
>python. The examples in the build tree work for C and C++, but there are
>some problems with the fortran examples in the build tree and no examples
>work yet in the install tree.  I expect all these problems are due to minor
>Mac OS X linking issues for shared libraries that still have to be worked
>out.  Or there may be some "obvious" CMake switch that needs to be turned on
>for the Mac OS X case.
>
>Finally, thanks to the efforts of Andrew Ross and yours truly, our CBS is
>working reliably on Linux with dynamic device drivers, shared libraries, and
>separate build tree.  Other variations on that platform need testing (e.g.,
>static devices, static libraries, source-tree build, separate prefixes for
>the install location and installed libraries location, cross-compilation,
>non-gcc compilers, non-standard build configurations, etc., etc.).
>Nevertheless, our CBS is looking good on Linux and both Andrew and I are
>impressed by its speed.
>  
>
Hopefully, we will be able to say the same of the Windows platform 
before long.

Regards,

Arjen