Re: [vile] Here is the file which demonstrates the insert bug

[Thomas Dickey]

On Thu, Apr 24, 2008 at 06:15:51AM -0400, Thomas Dickey wrote:
> On Thu, 24 Apr 2008, Chris G wrote:
>
>> On Wed, Apr 23, 2008 at 08:51:15PM -0400, Thomas Dickey wrote:
>>>
>>> I found another bug to investigate (don't know if it's related ;-).
>>>
>>> Using your test case, I looked to see if I could get valgrind to
>>> show me a problem (no...), and if I could get the 'working' mode
>>> to show that it was related (no...).
>>>
>>> But what I did was hold the key down that I was inserting, and just let it
>>> go for a while.  After about 50-60 screens shifted-right, the cursor and
>>> the insert-point stop tracking together (it continues to insert, but the
>>> two move at different rates).  That's about 3kb of inserted text.  But
>>> valgrind doesn't complain, and it's unrelated to 'working'.
>>>
>>> However, a quick check shows it doesn't happen in en_US locale - it's
>>> a problem with the en_US.UTF-8 locale.  I'll dig into that...
>>>
>> I've stirred up quite a can of worms here haven't I!  :-)
>
> I suppose so.  The one that I did fix was a bug that I'd like to have  
> caught before now (the reason for just focusing on bug-fixes for 9.7).
>
> The other I wouldn't have seen unless I was digging deeper to find
> something to reproduce from your report.  (It did seem like a lot of
> shift-right's, but I think I counted off 57-58 consistently - it has
> to be related to the specific data...)

hmm - I can see the cause.  I was running vile under valgrind,
and it made it run slowly enough that it stopped repainting lines
due to the way it handles typeahead.  The surprising thing is that
it continued to update the cursor movement - I'd not thought of that.

Without valgrind, I can't force it to run that slowly (though I suppose
I could add some time-delay to experiment with it ;-).

-- 
Thomas E. Dickey <address@hidden>
http://invisible-island.net
ftp://invisible-island.net

signature.asc
Description: Digital signature