[ADMB Users] Broken tpl2cpp in Windows

Johnoel Ancheta johnoel at hawaii.edu
Wed Apr 22 12:21:28 PDT 2009


Thanks Arni,

Created the issue

http://code.google.com/p/admb-project/issues/detail?id=59

Target for next Milestone3.

Johnoel

----- Original Message -----
From: Arni Magnusson <arnima at hafro.is>
Date: Wednesday, April 22, 2009 6:40 am
Subject: Re: Broken tpl2cpp in Windows
To: Johnoel Ancheta <johnoel at hawaii.edu>
Cc: users at admb-project.org

> I finally found the problem. The broken tpl2cpp in ADMB for 
> MinGW gets 
> created from the call:
> 
>    gcc -I. -DUX mylex.c -otpl2cpp
> 
> The -DUX should only be used when compiling mylex.c->tpl2cpp on 
> Linux. In 
> mylex.c, Dave Fournier defined a macro called UX that helps the 
> compiler 
> choose the right lines of code, depending on whether mylex.c is 
> being 
> compiled on Windows or Linux. By using -DUX it calls things 
> appropriate 
> for Linux, while bypassing the code intended for Windows. In 
> other words, 
> a healthy tpl2cpp for MinGW should be created with the call
> 
>    gcc -I. mylex.c -otpl2cpp
> 
> or better still
> 
>    gcc -s mylex.c -o tpl2cpp
> 
> for a leaner executable.
> 
> ---
> 
> The problem lies in mingw.mak, which currently calls 
> optg32-rh8-laplace.mak. As the name implies, the latter makefile 
> is 
> intended for Red Hat 8 Linux and is therefore not appropriate 
> for making 
> ADMB for Windows.
> 
> The solution is to edit mingw.mak so that it calls a makefile 
> intended for 
> making ADMB for Windows. This is a much better solution than 
> making ADMB 
> for Windows dependent on cat.exe. Dave Fournier already did the 
> hard work 
> creating the UX macro to make ADMB for Windows self-contained.
> 
> The same corrections need to be made for tpl2rem, which also 
> uses Red Hat 
> 8 Linux makefiles. It looks like the Windows makefiles will be 
> very 
> similar, except no -DUX.
> 
> Quack,
> 
> Arni
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.admb-project.org/pipermail/users/attachments/20090422/05f684cc/attachment.html>


More information about the Users mailing list