[Top][All Lists]

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

Re: [Bug-gnucap] Question about source in windows..

From: al davis
Subject: Re: [Bug-gnucap] Question about source in windows..
Date: Fri, 11 Jun 2004 09:27:16 -0400
User-agent: KMail/1.6.2

On Friday 11 June 2004 01:02 am, Sherif Abdul Kader wrote:
> Dear GNUCAP developers,
>  First of all, I would like to express my deep gratitude for
> your precious work. Although your software hasnt reached the
> stage where it is a REAL GNU software yet (its hasnt hit
> version 1.0 yet :-) ), its existence in the web is by itself
> a great advantage.
>  Second is my question. Are you planning to provide a MS VC++
> version of the source code?

Probably not.  If you want to do it and share it I will 
distribute it.

It is my policy to support standards and to provide portable 
code.  The code should compile on any C++ compiler that 
conforms to the official standard.  If it fails to compile on 
any standard C++ compiler, I consider that to be a serious bug.

When a compiler fails to meet the standard, then I must make a 
judgement of whether to work around it, or to just say "get a 
better compiler".  The double curlys on for loops are a 
consession to a bit of non-compliance with the standard in MS 

The real issue with MS products is a non-standard build process.  
Most others accept a "configure; make" build process, or 
preferably just "make".  If a system is only a little 
nonstandard, and doesn't do shell scripts, usually a few small 
hacks to the Makefile will do.  The used to work even with 
Microsoft and Borland.  Now, MS seems to have completely gone 
over to their IDE, which has big trouble doing anything 
unusual.  It also generates several huge binary files.

Usually, you can just load all the .h and .cc files and tell 
MSVC++ to compile them all.  You might need to remove one, or 
hack one.  You might need to change some options.

The real problem with the MS environment is when you need 
something unusual.  For Gnucap, this is in how the model 
compiler is used.  The correct way to build it is that you 
first build the model compiler, then use it to generate the C++ 
code for the models, then build the simulator.  The 
"configure;make" approach on a gnu system handles this well.  I 
never could figure out how to make MSVC do it.

If you must use MS windows, probably the best way to build GNU 
software there is "mingw".  http://www.mingw.org .  I believe 
that after installing mingw, the usual "configure;make" will 
build gnucap and most of the related tools easily.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]