4.3.0-dev-2 BUG: negative cursor position

4.3.0-dev-2 BUG: negative cursor position

[Jun T.]

Hi zsh-workers,

Is this the right place to report a zsh bug?

I tried 4.3.0-dev-2 with --enable-zsh-debug.

old-zsh% /usr/local/bin/zsh -f
new-zsh% bindkey -v
new-zsh% <ESC>kj

Here, <ESC> is to go into the vi-command mode.
Then I hit 'k' to go up in history stack; it worked as expected.
Then I hit 'j' to go down in the stack, but I get the following message:

BUG: negative cursor position

With gdb, I found that zlecs becomes negative at line 674 of zle_hist.c.
Replacing

    if (invicmdmode())

by

    if (zlecs > findbol() && invicmdmode())

"apparently" fixed the problems but I know NOTHING about the zsh source code.

Re: 4.3.0-dev-2 BUG: negative cursor position

[Wayne Davison]

On Mon, Dec 12, 2005 at 11:14:05PM +0900, Jun T. wrote:
> Is this the right place to report a zsh bug?

Yes, this is a very good place for that.

> With gdb, I found that zlecs becomes negative at line 674 of zle_hist.c.

Good catch -- that code is new for 4.3.x: it's a slightly optimized
version of some setline() code, but the check to see if zlecs was non-
zero was accidentally dropped from the algorithm.  I've checked in a
fix.  Note that your suggested fix will work just fine, but I ended up
using a slightly different check (taken from setline()) where it just
checks zlecs against 0 instead of calling findbol().

Thanks for the report,

..wayne..