Slash Boxes
NOTE: use Perl; is on undef hiatus. You can read content, but you can't post it. More info will be forthcoming forthcomingly.

All the Perl that's Practical to Extract and Report

use Perl Log In

Log In

[ Create a new account ]

autrijus (1505)

  (email not shown publicly)

Journal of autrijus (1505)

Sunday May 01, 2005
11:19 AM

Day 90: Pugs 6.2.2 released!

[ #24475 ]
On behalf of the Pugs team, I am delighted to announce the release of Pugs 6.2.2, the first version with a thread-safe, deadlock-free internal storage, based on software transactional memory (STM).

Also, the performance is much, much better with this release; it is at least 10x faster than the default 6.2.1 build, and more than 2x faster than optimized 6.2.1 build for common applications.

Because optimized build is now the default, compiling Pugs will take longer than before; please be patient. :-)

You can download Pugs from a nearby CPAN mirror, or from
SIZE = 786055
SHA1 = 0cda153c52590ea7ffc17b48195d413c8adc9e16

Much thanks to all the lambdacamels for making this release possible. It has been a wonderful three months working with you folks, and I look forward to many more months to come.


Changes for 6.2.2 - May 1, 2005

Pugs Internals

  • Many performance speed-ups, including:
    • New, much much faster implementation type for Pad
    • Replace Data.HashTable with STM Map
    • Restructured the Eval monad to be reusable on compilers and interpreters
    • Position handling and statement reduction logic re-written
    • Add strictness to core Monad and object structures
  • Functions declared as unary are now parsed as such
  • Global subroutine declaration is now visible in the whole file
  • Improved Parrot compiler backend support for mod_pugs
  • Much better error messages, tracking source file range for expressions
  • New loop control routines: next and redo
  • On Win32, -s and -z now work correctly
  • Optimized build is now default; use make unoptimized to turn it off
  • Pugs can now build on Cygwin
  • #line 123 and #line 123 "filename" both work
  • &infix:<%> raises trappable exception when modulus is zero
  • (undef, $x) = (1, 2) is now supported
  • @array.end implemented
  • BEGIN {...} blocks now work on both statement-level and term level
  • for @a { say } is now parsed as for @a -> $_ is rw { say }
  • for @a -> () { ... } now consumes one element at a time
  • my $var ::= expr works
  • print (4)+4 now prints 8, because the whitespace is no longer skipped
  • use Module; supported and happens at BEGIN time
  • |= ^= &= for junctions

Tests, Examples and Documentations

  • Many new tests and cleanup of older tests, we are now at 4600+ tests
  • API documentation available in docs/haddock/
  • Added an example of IRC bot into examples/network
  • Added simple POD parser into ext/Pod-Stream-Parser
  • Added support for generating Haskell documentation with make haddock
  • Added the tic-tac-toe game from perlmonks into examples/games/
  • Parallel make smoke when %ENV<PUGS_TESTS_CONCURRENT> is greater than 1
  • Patched local Test::Harness to display number of TODO tests
  • Progressive powerset generator from perlmonks added to examples/
  • Smoke testing tool now includes timing data in YAML harness data
  • German talk Perl 6, genau jetzt! finished

Bug Fixes

  • Constant references now automatically dereference when used in rvalue context
  • Junctive ^ should join junctions instead of autothread them
  • Lone block does not count as simple expression anymore
  • Repair single statements in shell to preserve lexical context
  • Repair the list context hinter
  • The dot in 123.ref is now parsed correctly as (123).ref
  • $_ in subroutine signatures is no longer slurpy by default
  • $x = \$x no longer causes an infinite loop
  • NaN calculations no longer causes infinite loop
  • last if foo now means last() if foo() instead of last(if(foo()))
  • one(1,1).pick bug fixed; it should return undef
  • splice([], 1) no longer causes division by zero errors
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
More | Login | Reply
Loading... please wait.