0 All those I don't know about! Let me know of any you find.
1 Not every version of make will build atlc. GNU make will do, which
is of course free. Sun's make will NOT. I think it is true to say
Sun's is the only version of make that is unsuitable, but if you find
another one, use GNU make, but please advise me of the version.
2.There are two test failures on IBM's AIX operating system, using IBM's
compiler version 6.0 when compiled multi-threaded I've no idea of the
source of these problems. Many pixels differ, but by a small amount. I
find it hard to believe this is rounding errors.
6d-create_and_check_50ohm_201Pixel_high_symmetrical_stripline.test
sometimes fails. Building multi-threaded tends to increase the
chances of a failure.
3 The file tools/src/locatediff does not work once compiled on AIX.
Despite althering the code to put a load of error checking in,
the thing still does not work. I've no real idea why, but since
it does not fulfill a major role, fixing this is non-urgent.
4.Data on machines is only gathered properly on a few machines (Solaris,
Tru64, AIX and HP-UX). On IRIX, Linux, *BSD, it is next to useless.
5.If there is a version of the gsl library installed, but it is too old
(< version 0.8), then the linker will link in -lgsl. However, there will
be no gsl functions called, so *some* linkers (well the only one known is
on SGI's IRIX platform, will generate the warning message
ld32: WARNING 84 : /usr/freeware/lib32/libgsl.so is not used for resolving any symbol.
This is harmless and may be ignored. The top-level configure.ac or the
m4/gsl.m4
files need sorting out to correct this
6.There seems to be problems on SPARC running old versions of Linux
(Redhat 6.2, kernel 2.2.14-5.0smp) and Debian, when configured for
multi-processor support. I'm tempted to believe this is a bug in the
thread support on old versions of Linux. It's fine on newer versions on a
PC.
7.The lack of any way of changing brightness of images without a
cpu-intensive re-run.
8.There are no manual pages for several programs
9.The error messages shown in the files docs/htlm-docs/errors.html are
wrong and were generated on an earlier release. However, it makes little
difference - the meaning should be clear.
10.HTML versions of the man pages have diagrams that are completely stupid. #
This is a problem with the program 'man2html' I'm using - I think so anyway.
11 Sometimes documention in one file is out of date with documention in
another. The web pages might not agree with those in docs/html-docs.
or README's contradict each other. Where possible I've tried to
remember to update the ChangeLog and add a date to a file, but this
does not always happen I'm afraid.
12 Sometimes the CVS is not as current as it should be.
Dr. David Kirkby, 30th Aug 2003 Re: atlc-4.4.0
Return to the atlc homepage