[Cl-perec-devel] Re: darcs patch: fix date-of-last-day-for-partial-date

[Attila Lendvai]

> Hello,

hi,

i hope you don't mind that i cc the list.

> > Do you think that your patch will correctly return:
> >
> > 2007-02-28 for 2007-02
> > 2007-03-31 for 2007-03
> > 2007-04-30 for 2007-04
>
> It does for me :)
>
>  PRC> (date-of-last-day-for-partial-date "2007-02")
>  "2007-02-28"
>  PRC> (date-of-last-day-for-partial-date "2007-03")
>  "2007-03-31"
>  PRC> (date-of-last-day-for-partial-date "2007-04")
>  "2007-04-30"
>
> Before it would return:
>
>  PRC> (date-of-last-day-for-partial-date "2007-02")
>  "2007-03-01"
>  PRC> (date-of-last-day-for-partial-date "2007-03")
>  "2007-04-01"
>  PRC> (date-of-last-day-for-partial-date "2007-04")
>  "2007-05-01"


i couldn't reproduce the breakage, but you changes look good and work
fine so i've pushed them. thanks!


> Even then, ASDF had two hiccups compiling everything:
>
> - a warning with closer-mop-utility-package (which I just "accepted"
>   and things seemed to be fine) and a


yeah, we also do that. we load it in the image so it was not annoying
enough to report/debug... :)


> - a error with alexandria's control-flow and not knowing what
>   MAKE-GENSYM-LIST is (which was an easy fix and I have a patch for).


i've pushed a fix to alexandria (missing dependency in the .asd)


> > Did the test suite more or less passed?
>
> Yes, after lots of setting up I got it where I pass nearly all the
> tests (+ a couple warnings I simply ignored):
>
> PRCT> (asdf:oos 'asdf:test-op 'cl-perec-test.postgresql)
>   ...
>   #<test-run 549 tests, 2105 assertions, 9 failures in 21.765 sec>
>
> The 9 failures were type inheritance errors. When you run (retest) you
> get 2 other extra errors.  I remember trying to look into this stuff, but
> I gave after finding myself running in circles. If you want more
> information (about the errors), let me know.


these are known issues. i guess we should add with-expected-failures
to stefil...


> > What kind of problems did you have with it so far?
>
> Mostly just those that I have mentioned. I haven't actually gotten
> around to *using* the thing. I was mostly evaluating my options for
> persistent storage, and although cl-perec looks *very* cool, I'm a
> little hesitant. Mostly because of:
>
> - the amount of privately maintained repositories scares me (it forces
>   me to wonder what would happen should the project, heaven forbid,
>   find itself abandoned)


most of them are basically the HEAD of the official repos, sometimes
lagging behind if there were non-compatible changes in the official
that we haven't followed yet, or holding one or two fixes that are
being sent to the maintainers.

i think the only repo holding a bigger change is postmodern: our
version uses local-time for the time related values and it wasn't yet
incorporated into the official (mostly because it removes support for
time intervals and the maintainer didn't know local-time enough at
that time to decide).


> - the fact that I don't predictable results with the test cases and
>   there are 9 failures


...but there are 2105 non-failing assertions... ;)

if we find something bogus, we usually add a failing test right on.


> - the fact that there's no documentation and some of the code I don't
>   quite understand yet.


i (we) can't promise anything about the documentation. i agree that
some bird's view doc is missing, but imho any other effort should be
put in the code and the test-suite instead.

i hope it answers some of the issues. otherwise feel free to ask!

happy hacking!

err, i mean merry christmas! :)

-- 
 attila