Print

Print


Hi Bruno,

 > > I took your package pushed it into my environment and first of all tried
 > > to 
 > > run our test suite against it. Of course that ended up in a huge number of
 > > failures.
 > 
 > I'm sorry about that. I'm couldn't get "make check" to work (on a
 > clean svn download), and I'm investigating that (I probably misplaced
 > some file, or permissions).

it is also quite likely that we have incorrectly built in some assumptions
that don't hold. perhaps we should offline try to sort out what goes wrong at
your end.


 > Right. I should have double-checked: I lost quite a few variants along
 > the way as well, and unfortunately, I was working from a version of
 > l3seq which was a few weeks old.

don't think that there have been much change here so that shouldn't be a
problem

 > > there are a number of advantages to this most importantly for me that we can
 > > run the test suite against the new code. So far I added
 > 
 > After a manual grep (more or less), I think we need to add the following.

I'll check what that gives me.

 > % Auxiliary functions (needed?).

aux functions should not be needed. the idea is that [aux] is internal to a
module, so if you replace the implementation no outside module should be
affected (in theory :-).

 > Hopefully that's ok. Of course, if other modules happen to try and
 > define their own seq functions using \seq_elt:w, then that's not going
 > to work.

yeah tough.

 > If possible, could you send me a log of those tests that fail? It
 > would help me find all the missing functions.

sure once I've run it

 > > So I propose we first make your implementation support the current
 > > interfaces (even if only by compatibility code) so that the files
 > > m3seq00... are properly processed and in parallel discuss its individual
 > > merrits and preferably we should at the same time write some additional
 > > tests for any functionality that has changed or is additionally there
 > > (Joseph will tell you that I'm pestering him about writing test cases in
 > > parallel wil code, but it really helps to nail down interfaces and ensure
 > > that they aren't violated later)
 > 
 > As I said, I can't run "make check" right now. Is it possible to run
 > single tests "by hand"?

in theory, something like

make checklvt F=m3seq003

should run the test on a single file, but you probably see the same issues
(whatever they are)

 > Then there is the issue of l3candidates, which I have not
 > re-programmed yet, as well as the quicksort defined in l3prg.dtx

one step at a time i'd say.

frank