[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: netCDF for Win32



> There was a typo in my description of the alleged MSOFT.MAK problem, I
> wrote 'LIB' where it should  have been '$(LIB)'.
>
> Profuse thanks to Glenn for his extremely useful MAKS.ZIP posting, and
> kudos to all the netCDF authors for the improvements!
>
> One issue that may be worth thinking about for the future is that the
> NETCDF.DLL itself could become a common file on Win32 machines, i.e.
> multiple applications from different sources will use it, and there may
> be issues with some of the compilation options, namely:
>
> 1) What C run-time library version should it use? (static vs DLL   &
>  Single-threaded vs multithreaded)       /MT, /MD, ... options
>
> 2) What structure member alignment options should it use? (1,2,4,8,16
> byte)        /Zp? options
>
> These options may not matter for the netCDF API, but we (Varian CSB) have
> had problems in our applications when mixing and matching these options.
> Since we have to read old data  files and want to run on data handling
> threads, we have standardized on 1 byte packing and multithreaded
> run-time DLL. I guess we could find out by building  nc_test and netcdf
> with different compile options and seeing what happens (???). The 32 bit
> Microsoft Visual Basic only supports 4 byte structure packing (Dilbert
> and Dogbert will have a field day if they find out...), so that may be a
> consideration for your guys, as interfacing  with Visual Basic is
> probably very desirable...

Part of our problem with this environment is we have very little
experience to base such choices upon.

If you have specific set of flags which you can recommend, I would
be happy to incorporate them.


> Thanks again,
>
> Jean-Louis Excoffier
>
> P.S. I'm very grateful for what you guys provided, but I think I should
> still let you know about my experience getting to the source code:
>
> Unless I missed something, the netCDF source is only provided a
> compressed TAR archive. Getting the actual source code on my PC was a
> fairly complex process, requiring the help of my Systems Administrator:
>
> 1) ftp the archive from my Windows 95 PC to a Unix machine
> 2) uncompress and tar it on a drive on the Unix system
> 3) using an NT machine (that can handle both Unix and Novell drives),
> xcopy the directory tree to a location on a Novell server
> 4) xcopy it down to my Windows 95 PC
>
> Some of the above steps could  have been bypassed, but my point is that
> even then many potential users could not deal with this... To gain wider
> acceptance  in the Windows/PC world, it would be useful to either post a
> ZIP archive (which is the PC standard), or to point to public domain
> versions of the TAR and Uncompress tools for the PC if they exist...

I'll provide it as a ZIP file next time. Thanks for the suggestion.

-glenn