Re: [Tcllib-devel] has anyone began testing tcllib cvs head and tcl8.5.4?

[Andreas K. <and...@ac...>]

>
> $ fgrep -c FAILED tcllib/unix/casrun.lwv
> 148
>
> Now, for each failure, there's two lines with FAILED in it. So 74
> test cases
> failed.

> It appears there are quite a large number of errors being
> generated when I run
> the tcllib test suite against tcl 8.5.4 built on sparc solaris.

> With this many errors, I don't have time, and you probalby don't have the
> energy, to handle that many bug reports.

Can you reduce it down to the names of the modules in which we have the
failures ?


> Here's a sampling of what I see:
>
> ==== do-2.3 wrong no of arguments FAILED
> ==== Contents of test case:
>
>     set res [catch do]
>     if {[string match  {no value given for parameter "body" to
> "do"*}  $::errorInfo]
>     } then {
> 	set ::errorInfo {wrong # args: should be "do body args"
>     while executing
> "do"}
>     }
>     list $res $::errorInfo
>
> ---- Result was:
> 1 {wrong # args: should be "do body ?arg ...?"
>     while executing
> "do"}
> ---- Result should have been (exact matching):
> 1 {wrong # args: should be "do body ..."
>     while executing
> "do"}
> ==== do-2.3 FAILED

Snit ... And it seems that the recent drive for 'consistent error messages'
(which started in the 8.5 series IIRC) has changed the details of the error
stack, causing many tests containing (partial) stack trac traces to now
fail.

We can likely fix this by "simply" updating the test cases. Maybe changing
to glob matching, to ignore the changing parts?



> This difference in error message appears to be the sort of thing many of
> the failures are.

Yes.

> Here's another type of error:
>
> ==== tree-tcl-4.4.17 error in walk FAILED

Also stacktrace

--
	Andreas Kupries <and...@Ac...>
	Developer @ http://www.ActiveState.com
	Tel: +1 778-786-1122