Re: [Nano-devel] updates, 1.3.9pre1 soon

[David Lawrence Ramsey]

Jordi Mallach wrote:

<snip>

>Sorry for not having replied to some of your last posts.

No problem.

>Debian bug #325065 describes a crash with long lines that sounds very
>much like what was fixed in 1.3.8. There's a file that makes it bomb,
>in theory, attached to the Debian bug, but I haven't been able to
>reproduce it.
>
>Have a look and tell me what you think about it.

Unfortunately, I can't reproduce it so far either, with xterm,
rxvt-unicode, or XFce's Terminal (which is at least based on vte).  Even
running it through valgrind doesn't raise any red flags about e.g.
reading past the end of an array; I just get a lot of "Conditional jump
or move depends on uninitialised value(s) ... (within /lib/ld-2.3.5.so)"
messages, much like those in Jan Jitse Venselaar's bug report earlier. 
There are some memory leak problems and related errors, but according to
valgrind, they occur inside functions called by regcomp().  Maybe this
is another glibc problem, but I'm not sure.

(Incidentally, I couldn't reproduce the crash in 1.3.7 involving
trailing tabs with the file supplied with it, either, but valgrind at
least gave me an error message when I tried replacing all spaces with
tabs in the file.  Doing that and similar things here doesn't seem to
affect anything.)

I did find and fix a display bug when I was trying to reproduce the
problem in fullscreen mode, due to an unhandled case in
display_string(), but that shouldn't make it crash.  Or, since 1.3.8-cvs
has a lot of internal changes, maybe that affects things.  Does the same
problem occur with current 1.3.8-cvs?

>Besides this, I'm all for 1.3.9pre1.

Oh good.