On Aug 16, 2011, at 4:39 PM, Neil Van Dyke wrote:

> 
> I don't know about this particular error, but there is a long history of 
> PLaneT packages giving install-time errors.  Sometimes (as Carl Eastlund 
> pointed out to me recently), it's because they're requiring a specific, older 
> version.  But there have been a few different causes.  And sometimes the 
> errors have been in PLaneT packages that are popular for other PLaneT 
> packages to depend upon.
> 
> I suspect a reason that these errors were allowed to persist so long is that 
> they're usually only noticeable at time of the PLaneT package install.  If 
> you re-run your program that "require"d the PLaneT package, after it failed 
> (or simply gave non-fatal error messages) due to PLaneT install-time 
> problems, you see no sign of trouble now that the PLaneT package is installed.
> 
> I think it's been mentioned before, but automated testing of PLaneT packages 
> by the PLaneT server/repository would be nice.  For starters, just see 
> whether they can be compiled and "require"d without error (in a sandbox).  
> Would be nice to test multiple Racket versions.  Later, as we get 
> standardized entry point for running test suites, the repository could even 
> detect regressions with new Racket versions and with combinations of 
> different versions of PLaneT packages on which a package depends.  
> Regressions in a supposedly backward-compatible version, or even any test 
> failure at all, could even trigger suspension of the version from new use 
> until a human approves it.  This is something doable on the repository side, 
> and something that simply won't get done if we're expecting individual 
> package authors to do it for their packages.

+1.

Also, I think it would be very helpful to try to compute the dependencies of a 
given planet package, and also a relative measure of its compile time.

Is the limiting factor here developer time or CPU time?

John


Attachment: smime.p7s
Description: S/MIME cryptographic signature

_________________________________________________
  For list-related administrative tasks:
  http://lists.racket-lang.org/listinfo/users

Reply via email to