Re: lynx-dev rc save bug

[Bela Lubkin]

Philip Webb wrote:

> > If you identify the hardware and OS, someone here ought to know:
> > `uname -a` is generally pretty useful.
>  
>  uname -a  gives IRIX 5.3, which i already mentioned in this thread.

It's not the sort of thing that people (other than the victim on the
spot) remember for more than 2 minutes.

> >> it's a 2-8 vs 2-8-1 problem, not a problem within recent devt versions.
> >> i can look thro' CHANGES, but people who were working on Options
> >> should be able to locate individual changes more quickly than me.
> > It would help even more if you could pinpoint which release broke it.
> 
> it takes  15 min  to configure & make each version: i'ld rather avoid it.

Your choice -- you're the one having the problem, you get to decide how
much effort you want to put towards resolving it.

> > I just ran:  truss -o lynx.truss lynx /tmp/foo
> > went to the option menu, checkmarked "save", hit "accept", and quit.
> 
> BTW what does  /tmp/foo  do?  foo  isn't a real file, surely?

I created an empty file, /tmp/foo, then used Lynx on it.  As I said,
that made a handy marker in the system call trace -- I could see when
Lynx called open("/tmp/foo").  Using an empty file minimized the
spurious system calls in the trace (no sequence of successful reads; no
socket operations etc. like there would have been if I used an http
URL).

> i suspect there is some other trace software on this system:
> can anyone suggest other names to try?

You got strace to compile -- try it.  Start by running it with no
options, it should give a usage message.  It should have a "-o" flag, so

  strace -o lynx.trace lynx /tmp/foo

should work.

> oh yes, how about Lynx debug?  should i recompile with that?

Doubt it'll help.

>Bela<