[racket-dev] A grab-bag of issues
At Thu, 4 Jul 2013 16:51:13 -0400, Sam Tobin-Hochstadt wrote:
> On Thu, Jul 4, 2013 at 1:05 PM, Matthew Flatt <mflatt at cs.utah.edu> wrote:
> > At Thu, 4 Jul 2013 12:52:10 -0400, Sam Tobin-Hochstadt wrote:
> >> 3. Rerunning `make` at the top-level loses the old set of PKGS
> >> specified. Should there be a `configure` setting for this to make it
> >> permanent?
> >
> > I agree that something should be done, but I'm not sure of the right
> > way to do it offhand. as you might imagine, it can be a little tricky
> > to find something that works with both Unix/Mac and Windows.
>
> Perhaps if there's a links file already, just keep it, unless PKGS is
> set? Or is that check the hard part?
You're right: the `PKGS' value is not needed until a Racket core build
is ready, so it's not difficult to record the value, etc.
Just keeping the links file doesn't work, since dependencies might
change. Keeping the currently installed non-auto links could work, but
I found that it's easy to hit Ctl-C at a point that loses the
configuration. So, there's a new "racket/etc/link-pkgs.rktd" file that
records your `PKGS' choice when the value of `PKGS' is not the default
value.
> >> 7. I think 'raco setup' should exit with a non-0 exit code when there
> >> are dependency problems.
> >
> > That sounds right.
>
> Does it already do this for other errors (such as documentation errors)?
I don't know what errors trigger a non-0 exit code now, and so that
needs to be reviewed in general.
> More issues:
>
> * Building on Travis with clang finds a lot of warnings, some of which
> look like bugs. See:
> https://travis-ci.org/samth/racket/jobs/8745819
I've pushed some repairs.
> * Similarly, that build shows that Racket detects itself as having
> many more CPUs than it actually has.
On Linux, the processor count comes from sysconf(_SC_NPROCESSORS_ONLN).
Should Racket use something else?
> * I think the FAQ and Future Plans sections of the pkg documentation
> are pretty out-of-date.
Yep.
> * If you start 'raco setup' and then immediately hit Ctrl-C, you get
> the message:
>
> uncaught exception: #<procedure>
>
> and setup doesn't actually stop.
Fixed.