Custom Query (1630 matches)

Adam Young pointed out an error in csc:

*** csc.scm.orig	2010-03-31 19:55:22.831032315 -0400
--- csc.scm	2010-03-31 19:56:47.520908976 -0400
***************
*** 253,259 ****
  (define linking-optimization-options default-linking-optimization-options)
  
  (define library-dir
!   (prefix "" "lib"
           (if host-mode
               INSTALL_LIB_HOME
               TARGET_LIB_HOME)) )
--- 253,259 ----
  (define linking-optimization-options default-linking-optimization-options)
  
  (define library-dir
!   (prefix "" "lib64"
           (if host-mode
               INSTALL_LIB_HOME
               TARGET_LIB_HOME)) )

It has to be checked, whether this is needed for all linux (or other UNIXoid) systems.

So rand() is completely subpar at generating random numbers in two ways:

1. Lower bits are not random. We already work around this in the standard manner by scaling the result of a floating-point division.

2. Nearby seeds don't affect the output much.

Due to # 2, on Mac OS X we may obtain the same initial random number for a minute or more, even though it is seeded with (current-seconds):

while :; do csi -p '(random 1000)'; sleep 5; done
734
734
734
734
734
209
209

We can fix the problem on OS X by using random(3) instead of rand(3), which is present on all BSDs and on Linux.

Internally, Linux uses random() when rand() is called and some BSDs seem to as well. However, there should be no harm anyway in applying the attached patch, which affects linux and bsd.

Stephen Eilert suggested the following:

-update-all: this would update all installed eggs to their newest versions.

-outdated: Instead of blindly updating, you'd be able to check which eggs are out-of-date

-list [name]: This would query the eggs available, preferably with a quick summary (similar to the current eggs page). The optional name argument would allow a filter.

-stale: Got the idea from rubygems. It would basically list the local eggs with their last atime. Might be helpful for when we reach a thousand gems ;)