[sword-devel] osis2mod import issue (and lack of regression tests)
dmsmith at crosswire.org
Sun May 31 04:19:17 MST 2009
On May 31, 2009, at 12:19 AM, Jonathan Marsden wrote:
> DM Smith wrote:
>> I had accidentally reverted Ben's change and have just put it back.
> Bad timing... we uploaded SWORD 1.6.0 packages to Debian a few hours
> ago, with my patch that included all the fixes up to r2420... looks
> we should do another package in a coupld of weeks with this fix in
> it too...
> Incidentally, while I remember... it would have been easier for me if
> you'd committed the debugging enhancements to svn as a separate commit
> from the bug fixes... I had to manually separate them out, since I saw
> the debug changes as an enhancement, not a bug fix.
Good point. I'll try to keep each commit to either enhancements or bug
fixes and one "issue" at a time.
> I've grabbed this latest (r2421) bugfix, and a change to re-include
> tcl biblebot (in the examples only, not as an executable), and
> started a
> new packaging branch we can put out as 1.6.0-2 when the time is right
> for that.
> BTW, this kind of issue *really* demonstrates the need for regression
> tests for the utilities, doesn't it :)
> I'd love to propose utility regression tests (and 100% working test
> suites!) as new features to aim for in SWORD 1.6.1 :)
I'm working on a regression for osis2mod. I have a few different xml
files I use to test how osis2mod handles valid OSIS constructs that
have been problematic in the past.
If this were set up as a regression, it would have caught the problem.
More information about the sword-devel