[R] R 1.7.0 (Windows) Crashes After using "Install Package from Cran option" within

[Joel Kincaid]

dmurdoch at pair.com wrote:
> On Thu, 24 Apr 2003 19:56:37 -0400, you wrote:
> 
> 
>>I should have added to my previous email that the work around mentioned 
>>at 2.17(a) does not correct the problem.....
> 
> 
> What is 2.17(a)??
Sorry: in the R faq for windows port installed with R1.7.0 and reaD in a 
browser generated from help.start() in Rterm is sectioned as
2.17 -- downloading files,
two options exist--a and b. I've tried (a) (using internet2.dll) and I'm 
  still trying to figure out (b).

>>When I search() in Rterm methods is not present in the path and 
>>furthermore I get messages to the effect of 'methods' not loaded:
>>
>>When I first extecuted demo() I get
>>"warning message:
>>package methods in options '(defaultpackages)' not found."
> That's a sign that there is something wrong with your installation.
> If you look in the R home directory, do you see a subdirectory called
> library/methods?
YES:    indeed if i call 'library(methods)' the package is loaded. If I 
then call demon() again -- I get no error message.

>>>After installing the program with no hitch  I used the the 'Install 
>>>Packages from Cran' optinon in the drop down menu from Rgui. (This may 
>>>be bad form, if so let me know): I selected all paackages to install and 
>>>everything was fine until the 'dse' package was downloading. 
>>
> 
> It's probably not a good idea to install all of the packages that are
> available 
>  - it uses a lot of bandwidth to CRAN (unless you've chosen a mirror
> closer to you)
>  - it uses a lot of your disk space
>  - it gives you lots of hits whenever you do a search for help -
> you'll find way more than is useful.
> 
> But it should work (except that it will fail on dse, unless you're
> using the patched version of 1.7.0, along with a not-yet-available
> repackaging of dse).  However, the fact that methods is missing from
> your installation indicates you've got serious problems.
> 
Yes I used a mirror and I'm a bit of a packrat (and the marginal cost 
---to me--of an additional MB of storage is zero (and this is true 
unforuntaly of bandwidth too).

> Most of this information isn't any use.  If you have DrWatson or a
> similar crash analyzer, there may be a few lines of useful information
> (e.g. the location of the instruction pointer at the time of the
> crash), but mostly it's not all that helpful.  Easily executed
> instructions for reproducing the crash are the most useful information
> you can give.
> 
> Duncan Murdoch

I'll try to dig up dr watson and see what I get. But the odd thing is 
that Rgui works when intally installed: it just failed after I try 
grabbing all of the packages.

In terms of reproducing the results (I've done this twice with the same 
results)
1. download and install the 21 MB file (from a mirror close to you),
2. bring Rgui up
3. ---works like a charm
4. gunk up the works by trying to (from the menu) install all pacakges.
5. wait for this to hang.
6. upon hanging, exit Rgui
7. try to start-up
8. finish installing all pacakges (again from menu)
9. exit Rgui
10. try to restart Rgui---it should then fail to load.

can the process of hanging and then using having give the three finger 
salute be corrupting some file????

thanks for taking the time to look into this,
joel