RE: setup.exe 2.218.2.8/9 broken

["Robert Collins" <robert dot collins at itdomain dot com dot au>]

> -----Original Message-----
> From: Harig, Mark A. [mailto:maharig@idirect.net] 
> Sent: Saturday, May 18, 2002 4:07 AM
> 
> It's the "do not choose" portion of this solution that I hope 
> setup.exe
> would avoid because it isn't paying attention to Murphy's 
> Law.  The way
> setup.exe runs now there are (at least) two possible sources 
> of errors:
> 
>   1) setup.exe's setup.ini has become corrupted.  This is well handled
> by setup.exe with its display of parsing errors.
> 
>   2) The user may have selected a Local Package Directory 
> that contains
> non-setup.exe setup.ini files.  The message that is reported for this
> user error (that is, 'user selected a Local Package Directory that has
> non-setup.exe package files in it') is, unfortunately, the 
> same message
> as that used to report problem 1, above.

This is because setup cannot tell the difference. Can it be taught to
tell the difference? Possibly, but I won't be doing the teaching.
 
>      This can be doubly confusing because a user can run setup.exe
> successfully for a long time, and then find that it stops 
> working due to
> mysterious parsing errors because s/he has installed some 
> other package
> ("I'm keeping all of my installations in a single, separate directory
> tree").

I keep all mine in a single separate directory tree too. This
functionality will never go away. It's keeping setup's 'local package
dir' at a logical node in the tree rather than a leaf that is the
problem.
 
>      So, even if we add the text you suggested telling the user about
> the rules for 'Local Package Directory', setup.exe should report the
> error better (i.e., not reuse the error processing method of different
> kind of error) when the user doesn't follow the rules.

It's the same sort of error to setup. I'll accept a patch that can
-accurately- differentiate between the problems. Don't diff against the
distributed source, diff against CVS for this please, as the HEAD parser
is somewhat...different.

Rob

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/