gEDA-user: Compiling MMI (electromagnetic simulator)

[Karel Kulhavý <clock@twibright.com>]

Hello

MMI promises to compute distributed paramteres of copper strips on
PCB.

I chewed throught the latex2html stumbling block and hit another one
-- mmi (aka tnt - I don't know why there is a naming schizophreny)
installed without any complaints, but refuses to run:

clock@beton:~$ tnt
Error in startup script: can't find package Itcl
    while executing
"package require Itcl"
    (file "/usr/local/bin/tnt.tcl" line 114)
    invoked from within
"source [file join [file dirname [info script]] tnt.tcl]"
    (file "/usr/local/bin/tnt" line 19)

I determined I need some package called "itcl" (don't know what it does,
but it isn't mentioned in README) so I stuck it into google and found itcl
homepage:

http://incrtcl.sourceforge.net/itcl/

There are several versions where I randomly chosen the latest one because
don't have any clue what version I should install. Got 3.2.1 from
http://sourceforge.net/project/showfiles.php?group_id=13244

The ./configure process crashed and the README showed to be ambiguous:
"
      If your Tcl installation is sitting somewhere other than right
      next to this package, you may have to tell configure where to
      find it:

        ./configure --with-tcl=/usr/local/tcl/lib
"
There were two candidates for this ./configure flag:
clock@beton:~/itcl3.2.1$ find /usr/lib -maxdepth 1 -type d -name '*tcl*'
/usr/lib/tcl8.4
/usr/lib/tcllib1.4

The right one is none of them, it's --with-tcl=/usr/lib
With this the configure process didn't crassh on error, just issued a warning
about missing tk. I got an idea to put also --with-tk=/usr/lib there that
cause the ./configure script to run without any complaints. A bugreport
resulted, of course.

This is probably a result of my not being a tcl/tk guru. However I think
the trace simulator will be run also by other people that aren't tcl/tk
gurus.

After that, however, the make process failed with 2 screens of syntax error
messages of this kind:
cc -DVERSION=\"3.2\" -DUSE_TCL_STUBS=1       -DITCL_LIBRARY=\"/usr/local/lib/itcl3.2\" -I/home/clock/tcl8.4.4/generic -I/home/clock/tcl8.4.4/unix -I"./generic" -I"./unix"    -O -D__NO_STRING_INLINES -D__NO_MATH_INLINES  -fPIC -c `echo ./generic/itclStubInit.c` -o itclStubInit.o
In file included from generic/itclStubInit.c:14:
generic/itclInt.h:53:20: tclInt.h: No such file or directory
In file included from generic/itclStubInit.c:14:
generic/itclStubInit.c:142: warning: excess elements in scalar initializer
generic/itclStubInit.c:142: warning: (near initialization for `itclIntStubs')
generic/itclStubInit.c:143: warning: excess elements in scalar initializer
generic/itclStubInit.c:143: warning: (near initialization for `itclIntStubs')

Another bugreport resulted together with realization that the author of the
software is in an out-of-office state.

Don't you please have someone experienced in making these things work?

I am lured by an idea of computing how many picoforad 1cm^2 of PCB trace on my
1.5mm FR-4 board makes. 

I didn't do anything deliberate to break my installation (smashing a hammer
into my harddisk or randomly deleting files) aside from installing programs
according to their README's. I have an university degree from computer systems,
operating systems, compilers and networks, however am still clueless here.

I had to pass exams from how good compilers and operating systems are being
programmed and networks built. However they didn't teach us there even a single
lesson how to fix badly written programs. 

I think that a person that isn't involved in UNIX internals at all, has a
technical degree, has part of his life been designing PCB's and now wants
to compute a couple of traces will be at best in the same position as me.

I think this is the main problem of some contemporary free gEDA tools: they
(sometimes, often) don't work when you follow the README.

Cl<