[R] Rgui.exe fix for "console not found" buffer overflow problem

Duncan Murdoch murdoch at stats.uwo.ca
Sun Oct 29 16:25:19 CET 2006


On 10/29/2006 9:55 AM, Gabor Grothendieck wrote:
> On 10/29/06, Gabor Grothendieck <ggrothendieck at gmail.com> wrote:
>> On 10/29/06, Duncan Murdoch <murdoch at stats.uwo.ca> wrote:
>>> On 10/29/2006 8:23 AM, Gabor Grothendieck wrote:
>>>> On 10/29/06, Duncan Murdoch <murdoch at stats.uwo.ca> wrote:
>>>>> On 10/29/2006 5:52 AM, Duncan Murdoch wrote:
>>>>>> On 10/29/2006 1:12 AM, Jesse Albert Canchola wrote:
>>>>>>> Here's how the problem started.  In Rgui.exe, I attempted to get more
>>>>>>> visible output in the buffer (I wanted to be able to scroll up the to the
>>>>>>> top and see the variable names, etc.), so I put the max of all 9's in the
>>>>>>>
>>>>>>> Rgui.exe
>>>>>>>
>>>>>>> pull-down menus
>>>>>>>
>>>>>>> Edit > Gui Preferences... > buffer bytes 9999999999999
>>>>>>>
>>>>>>> and
>>>>>>>
>>>>>>>> lines 9999999999999999999
>>>>>>> not realizing this would effectively break the Rgui.exe and no matter what
>>>>>>> you did to clean R out of your machine and re-install it, you would get
>>>>>>> the same thing (with both v2.3.1 and v2.4.0).  After a couple of weeks of
>>>>>>> using Rterm.exe and meeting deadlines, I finally had time to wade through
>>>>>>> the vague references online regarding where the problem might be.
>>>>>> Thanks for the report.  I'll take a look and see if there's a more
>>>>>> friendly way to handle this.
>>>>>>
>>>>>> Generally starting with the --vanilla option is a good idea to help with
>>>>>> diagnosing startup problems.  If they go away, you've got a problem in
>>>>>> one of your startup files, and ?Startup tells you how to trace through
>>>>>> the startup process.
>>>>> Whoops, I forgot that --vanilla doesn't stop an attempt to load
>>>>> Rconsole.  I'll take a look, and see if perhaps it should.
>>>> Also it might be useful if the Edit GUI screen listed the location of
>>>> the Rconsole
>>>> file it is using so its not so mysterious.
>>> The problem is that that dialog is not using an Rconsole file, it's
>>> using internal values.  Those may have been loaded from an Rconsole, or
>>> they may just be defaults.
>>>
>>> But you're right, in the case where they were loaded from a file, it
>>> would be nice if it were easier to find out what that file is.  (It
>>> wouldn't help in the situation here, because you can't get to that
>>> dialog:  R won't start at all if someone has messed up their Rconsole
>>> file sufficiently.  If --vanilla prevented the Rconsole file from
>>> loading, again it wouldn't help, because the values would be defaults,
>>> not those from the file.)
>>>
>>> Duncan Murdoch
>>>
>> There is a "Load" button on the Edit | GUI Preferences screen.
>> Does that load the values from the Rconsole file?  If so, then perhaps
>> it could be expanded to show which Rconsole file it would load
>> from if it were pressed.
>>
> 
> I guess I should have tried pressing it before I posted but if you do
> press it then it comes up with a browser window pointed at the
> Rconsole file and giving one a chance to change which file it loads
> before actually loading it so this seems like an easy way to find the
> Rconsole file that is actually being used.

Right.  This would be more obvious if I'd followed the standard Windows
convention and labelled it as "Load..." instead of "Load".  The dots
indicate that you'll get another dialog and opportunity to cancel.  I'll
change that.

Duncan Murdoch



More information about the R-help mailing list