Re: [Monotone-devel] [PATCH] new selectors: "earlier or equal" and ?later"

[rghetta]

Nathaniel Smith wrote:
> > +static void
> > +sqlite3_normdate_fn(sqlite3_context *f, int nargs, sqlite3_value ** args)
> What is this normdate stuff for?  Sqlite allows <= on strings, and it
> compares as "strcmp", which seems it should work perfectly well for
> normalized ISO dates?
Actually, my first try did just that, but I had problems getting the 
correct ranges to work.
Converting to double did the trick, and while is less efficient than 
comparing directly the strings, I wasn't very concerned, given the 
intended use of this function.
If you want, I can make another try using only strings; perhaps it was 
just some silly mistake on my part (coding at night has its drawbacks :-)

> > -    if (str.find_first_not_of(constants::legal_id_bytes) == std::string::npos
> > -       && str.size() == constants::idlen)
> > +    if (str.find_first_not_of (constants::legal_id_bytes) ==
> > +                               std::string::npos && str.size () == 
> > constants::idlen)
>
>    ^^ these are literal tabs.  Please do not submit patches containing
> tab characters?
> (Among the other reasons to not use tabs, there's a fair amount of
> noise in this patch from just this.)
I agree. I just hate tab intendenting. Unfortunately part of this patch 
was developed on a laptop and I forgot to configure it properly.
My apologies.

> If you want to test the behavior of the default hooks, I suggest using
> RAW_MONOTONE instead of MONOTONE (there are several tests that do this
> already).  This is much better than copying the hook into the
> testsuite, because if the standard version in monotone changes, then
> we will be testing the version in the testsuite rather than the
> version in monotone.  Which would make the test kind of pointless :-).
Oops, I haven't noticed RAW_MONOTONE. I'll look at that.
BTW there's a test macros documentation somewhere (other than 
test/README) ?

Riccardo