Dr. Volker Zell wrote:
Builds fine from source when using your patched cygport file, packaging and
setup.hint look good
Thanks, Dr. Zell. I'm ready to upload inetutils-1.5-2 as a *test*
release, but I have one remaining concern: I still haven't figured out
why I was getting setup.exe failures.
As a reminder, the situation was:
(1) I created a local, incomplete cygwin "mirror", with my new
inetutils packages
(2) generated a setup.ini using genini. This setup.ini looks good as
far as I can tell.
(3) running setup.exe with that directory, my new package does not
show up (at first). If I clock 'Back' a few times, so that the same
setup.ini is re-read, then it DOES show up in the chooser. But when I
'choose' it, nothing is installed.
I tried to debug it but ran out of time -- I also discovered a few
null-pointer accesses that occur in this situation, but only if DEBUG is
defined [*]. I *hope* the non-DEBUG subobptimal behavior is an artifact
of my incomplete local mirror.
But at this time, all I can do is upload the new packages, and hope for
the best. So, I want to stand ready to pull them, if they do induce
problems when included as part of a full mirror.
[*] Because -DDEBUG/#define DEBUG is not a normal part of setup.exe's
"debug" builds (which only trigger -g and -O0), it looks like some of
the DEBUG code has bitrotted.
--
Chuck
