Editing Pugs (compiler) (section)

==Development model==
Several factors have been suggested as reasons for Pugs's progress:
* Haskell's [[static typing]] can make it easier for program bugs to be detected at compile time. Haskell code is also often thought to be concise. The Parsec library [https://web.archive.org/web/20060807052639/http://www.cs.uu.nl/~daan/parsec.html], a [[Monad (functional programming)|monadic]] [[combinatorial]] parser written fully in Haskell, simplifies parsing. Because Haskell is a [[purely functional programming]] language, making the functional code interact with the real world ([[input/output]] and time-driven environment) requires thought. To achieve this, Pugs makes extensive use of [[Monad (functional programming)|monads]].
* Pugs's use of [[test-driven development]] methodology (a tenet of [[Extreme Programming]]). This methodology dictates that every module should have test code, even before the modules are implemented. Advocates of this methodology argue that it improves [[software quality]]. However, the project often silenced failed regression tests before releases, removing much of the benefit of test-driven development.{{Citation needed|date=May 2008}}
* Tang's liberal granting of the [[commit bit]]. Pugs development is currently based around a [[Apache Subversion|Subversion]] repository, and access is freely given, especially to people wishing to write tests. Because of this, a huge library of tests has accumulated. Other Perl 6 implementations rely on many tests developed for Pugs as an executable specification for Perl 6.
* Tang's communication style; her journal (linked below) attracted many people to the project.

Despite these factors, progress on the Haskell implementation stalled in late 2006, as personal issues kept Audrey from devoting as much time to the project as she had in 2005.

Many Pugs contributors have since moved on to implement Perl6-inspired systems as CPAN modules on Perl 5, such as the [[Moose (Perl)|Moose]] project.