Help review the error of fwts-test

[Colin Ian King]

On 19/06/12 08:49, Keng-Yu Lin wrote:
> Hi all,
>    All the patches are cleared with V0.25.03 tag in the git and
> packages uploaded to the Unstable Crack PPA.
>
>    But we still have a few fwts-test errors to conquer (I attached my
> own for reference).  The fwts-test git tree is also up-to-date (but
> without the tag because I expect some more patches for these errors).
>    Please help with fixing this error and fire the patches for review. \o/
>
>    Thanks,
> -kengyu
>
>
>
OK - there are a bunch of errors here because of the way logging line 
numbers are handled.  In the previous version, I bumped the log line 
number on each new line output *after* it was formatted to fit into the 
line width.  So text wider than the tty width would be broken into 
multiple lines, each one would count for a line number.

With the new logging, the notion of a log line was more fuzzy, since 
each output formatting engine can chose to break or not break lines to 
how it chooses. For example, html dynamically does test breaking when 
one re-sizes the browser page width.  So, I changed the notion of a line 
number to Nth fwts_log_* output line. If a line gets broken or 
reformatted it can have multiple lines mapping to this Nth line.

So, I've really re-defined what "line number" means. Originally it mean 
the Nth line in the log.  Now it means the Nth line output from fwts. 
Thus all Nth lines match on all output log back-ends, but one may see 
one Nth line map to one or more real physical lines in the log.

I'll make some adjustments to handle this in fwts and fwts-tests.

Colin