R devel - Nov 2005 - R: bug in windows GUI/script editor (PR#8288)

Hi Duncan,
I am running R in MDI mode (with localisation not installed because I need to
use RWinEdt), on an Italian version of Windows 2000 Pro.
I just logged in today and tried to reproduce the bug: the first three times all
was fine, the fourth one I got the bug, and finally noticed one detail: the bug
appears when I open an R script (I tried several) which causes the "part of
the scrollbar which one can drag" in the internal editor window to be
resized - i.e.:
- the editor window opens up (not maximised) with the vertical scrollbar about
1/3 of the window height
- the scrollbar gets resized to about half of what it was (without my
intervention)
- now I just close the window, click on menu File and there I get bitten

May it be the file size? 'wc' gives for a couple of cases in which I
have problems:
   333   1097  14681 Analisi-std-2.R
   337   1391  10532 R-riop-distributions.R

I also tried with plain R (i.e., my .Rprofile is not loaded): same behaviour.

Hope this helps; let me know if and what other details I can provide.
Regards,
  Roberto

> -----Messaggio originale-----
> Da: Duncan Murdoch [mailto:murdoch@stats.uwo.ca]
> Inviato: martedì 8 novembre 2005 19:00
> A: Ugoccioni Roberto
> Cc: R-bugs@biostat.ku.dk
> Oggetto: Re: [Rd] bug in windows GUI/script editor (PR#8288)
> 
> 
> On 11/8/2005 11:51 AM, roberto.ugoccioni@sanpaoloimi.com wrote:
> > Full_Name: Roberto Ugoccioni
> > Version: 2.2.0
> > OS: Windows 2000
> > Submission from: (NULL) (193.203.232.5)
> > 
> > 
> > Running Windows 2000 Professional, all patches up to nov 8, 2005.
> > 
> > 
> >>version
> > 
> >          _              
> > platform i386-pc-mingw32
> > arch     i386           
> > os       mingw32        
> > system   i386, mingw32  
> > status                  
> > major    2              
> > minor    2.0            
> > year     2005           
> > month    10             
> > day      06             
> > svn rev  35749          
> > language R  
> > 
> > How to reproduce the bug:
> > 
> > 1. launch Rgui.exe
> > 2. menu File->open script
> > 3. close editor clicking on X
> > 4. clicking menu File now generates the fatal error (omitting memory
> > addresses):
> > 
> >  An instruction referred to a memory location which could 
> not be "read"
> > 
> > Clicking on OK in the error message window causes R to 
> consume 100% CPU and not
> > to respond - must be terminated from task manager.
> 
> I don't see this, but it sounds like something that was fixed 
> before the 
> release of 2.2.0, so maybe there's another way to generate 
> the same problem.
> 
> Could you give more detail:
> 
>   - are you running in the default MDI mode (one big window 
> containing 
> the console, editor, etc.) or SDI mode (separate windows)?
> 
>   - does it matter what was in the file you opened?
> 
> Duncan Murdoch
> 

Il  contenuto  e  gli allegati  di questo  messaggio  sono  strettamente
confidenziali,  e ne sono vietati la diffusione e l'uso non autorizzato.

Le  opinioni  ivi  eventualmente  espresse sono  quelle  dell'autore: di
conseguenza  il  messaggio  non  costituisce  impegno  contrattuale  tra
il Gruppo Sanpaolo  ed  il  destinatario,   e  la  banca  non  assume  alcuna
responsabilita' riguardo ai contenuti del testo e dei relativi allegati,
ne' per eventuali intercettazioni, modifiche o danneggiamenti. 
 
Qualora il presente messaggio Le fosse pervenuto per errore,  Le saremmo
grati  se lo  distruggesse e,  via e-mail,  ce ne comunicasse  l' errata
ricezione all'indirizzo postmaster@sanpaoloimi.com.


This e-mail (and any attachment(s)) is strictly confidential and for use
only by intended recipient(s).  Any opinions therein expressed are those
of the author.  Therefore  its content  doesn't represent any commitment
between   Sanpaolo Group  and  the  recipient(s)   and   no  liability  or
responsibility  is  accepted  by  Sanpaolo Group  for  the above mentioned
content.

Sanpaolo IMI S.p.A. is a Bank authorised by Banca d'Italia; Sanpaolo IMI 
S.p.A., London Branch, is regulated by the Financial Services  Authority
for the conduct of investment business in the UK.

If    you   are   not   an   intended   recipient(s),    please   notify
postmaster@sanpaoloimi.com promptly and destroy this message.

On 11/10/2005 3:02 AM, roberto.ugoccioni at sanpaoloimi.com
wrote:
> SGkgRHVuY2FuLA0KSSBhbSBydW5uaW5nIFIgaW4gTURJIG1vZGUgKHdpdGggbG9jYWxpc2F0
Roberto:

Something went wrong with your posting -- it came out completely in 
binary.

I think I have found what was wrong with the editor windows.  There were 
two problems:

  - a variable was sometimes uninitialized.  I've fixed this in R-devel 
(though it won't be downloadable for a while).  I had hoped this was 
enough, but it wasn't.

  - the script editor window duplicates some menu items from the main 
window, but unfortunately in doing so it overwrites some pointers to 
those items.  When the script editor is closed, the pointers become 
invalid, and you may crash.

The second fix will require more extensive work, but I should get it 
done today.

Duncan Murdoch

On 11/10/2005 8:58 AM, Duncan Murdoch wrote:
> On 11/10/2005 3:02 AM, roberto.ugoccioni at sanpaoloimi.com wrote:
> 
>>SGkgRHVuY2FuLA0KSSBhbSBydW5uaW5nIFIgaW4gTURJIG1vZGUgKHdpdGggbG9jYWxpc2F0
> 
> 
> Roberto:
> 
> Something went wrong with your posting -- it came out completely in 
> binary.
> 
> I think I have found what was wrong with the editor windows.  There were 
> two problems:
> 
>   - a variable was sometimes uninitialized.  I've fixed this in R-devel
> (though it won't be downloadable for a while).  I had hoped this was 
> enough, but it wasn't.
> 
>   - the script editor window duplicates some menu items from the main 
> window, but unfortunately in doing so it overwrites some pointers to 
> those items.  When the script editor is closed, the pointers become 
> invalid, and you may crash.
> 
> The second fix will require more extensive work, but I should get it 
> done today.
Now committed to both R-patched and R-devel.  Will show up in the 
binaries on CRAN in due time.

Duncan Murdoch