Comments on: Mandatory versus optional ISO Prolog standards http://blog.logtalk.org/2009/10/09/mandatory-versus-optional-iso-prolog-standards/ Logtalk development happy moments, hurdles, and dreams Thu, 17 Jun 2010 00:23:27 +0000 hourly 1 http://wordpress.org/?v=3.0.1 By: Paulo Moura http://blog.logtalk.org/2009/10/09/mandatory-versus-optional-iso-prolog-standards/comment-page-1/#comment-682 Paulo Moura Fri, 11 Dec 2009 15:59:30 +0000 http://blog.logtalk.org/?p=395#comment-682 I don't advocate dropping modules from Prolog. I took great pains to enable Logtalk to compile modules as objects. In doing so, I looked for common core functionality, based on current practice. What I advocate is that this common core functionality should be a standard for ISO Prolog modules, ditching the currently approved but mainly ignored standard. Nevertheless, Logtalk objects do provide a replacement for modules *as* implemented in current Prolog compilers. It seems that you're talking instead of modules as functional equivalent to object namespaces. For big applications that can be a desirable feature. I toyed around implementing it but never found a real case that truly justified the added constructs. I don’t advocate dropping modules from Prolog. I took great pains to enable Logtalk to compile modules as objects. In doing so, I looked for common core functionality, based on current practice. What I advocate is that this common core functionality should be a standard for ISO Prolog modules, ditching the currently approved but mainly ignored standard.

Nevertheless, Logtalk objects do provide a replacement for modules *as* implemented in current Prolog compilers. It seems that you’re talking instead of modules as functional equivalent to object namespaces. For big applications that can be a desirable feature. I toyed around implementing it but never found a real case that truly justified the added constructs.

]]> By: Alan Baljeu http://blog.logtalk.org/2009/10/09/mandatory-versus-optional-iso-prolog-standards/comment-page-1/#comment-681 Alan Baljeu Fri, 11 Dec 2009 15:29:25 +0000 http://blog.logtalk.org/?p=395#comment-681 The purpose of ISO is to enable portable code. Real-world code needs modules. If modules aren't in the core standard, ISO prolog will be worse off. However, if as Parker suggests, an alternative means of organizing code is instituted, that could be fine: Make something new that can work in any Prolog; and deprecate modules, giving the world a gradual migration path. But objects aren't a replacement for modules. Modern object oriented languages also have namespaces. The purpose of ISO is to enable portable code. Real-world code needs modules. If modules aren’t in the core standard, ISO prolog will be worse off.

However, if as Parker suggests, an alternative means of organizing code is instituted, that could be fine: Make something new that can work in any Prolog; and deprecate modules, giving the world a gradual migration path.

But objects aren’t a replacement for modules. Modern object oriented languages also have namespaces.

]]> By: Parker http://blog.logtalk.org/2009/10/09/mandatory-versus-optional-iso-prolog-standards/comment-page-1/#comment-133 Parker Fri, 09 Oct 2009 15:59:25 +0000 http://blog.logtalk.org/?p=395#comment-133 It makes sense to push for the mandatory core ISO without modules given the current situation with modules. By hindering the core ISO, committee members opposing this step certainly don't represent my interests so I hope they are representing interests of the wider community and not just their own. Slightly off-topic: Paulo, given Logtalk is a core element of your research activities I sometimes fail to grasp why you are flogging the modules dead horse. Even if modules were completely 100% repaired, standardised, and uniformly adopted by all Prolog implementers, they'd still be "just" modules. Most modern programming languages used for programming-in-the-large have adopted object-orientation. Surely the way forward is not to try to fix something that will be mediocre at best, but to have better and better systems for logic-programming-in-the-large. It makes sense to push for the mandatory core ISO without modules given the current situation with modules. By hindering the core ISO, committee members opposing this step certainly don’t represent my interests so I hope they are representing interests of the wider community and not just their own.

Slightly off-topic: Paulo, given Logtalk is a core element of your research activities I sometimes fail to grasp why you are flogging the modules dead horse. Even if modules were completely 100% repaired, standardised, and uniformly adopted by all Prolog implementers, they’d still be “just” modules. Most modern programming languages used for programming-in-the-large have adopted object-orientation. Surely the way forward is not to try to fix something that will be mediocre at best, but to have better and better systems for logic-programming-in-the-large.

]]>