|
From: | Massimiliano Gubinelli |
Subject: | Re: [Texmacs-dev] plugin for Haskell |
Date: | Fri, 22 Jun 2018 18:31:32 +0200 |
No, I’m quite busy lately. I went a long way to switch TeXmacs to chibi but then I realised that a standard r7rs implementation is not flexible enough to handle all the behaviours that TeXmacs' scheme needs and that Guile allows. s7 seems a more pragmatic scheme where maybe one can make things work. There are two main difficulty in porting TeXmacs’ scheme to other interpreters than Guile. All these are mainly related to tm-define and tm-define-macro which implement TeXmacs overloading system. 1) modules can refer to identifier which will be tm-defined later and in other modules without any specific forward declaration of which these identifier be. Chibi for example assing “undefined” to these indentifier since they are not in scope at the moment the module is read. Even if the identifier become available later on this binding will not be modified and that is too rigid. If one would like to stick to a standard r7rs implementation maybe one should add forward declaration to the code, which would mean go on an modify much of the existing scheme files…. 2) r7rs schemes do not have a flexible enough module system, in particular one need the possibility to “open” a module and add new idenfitier or macro to the list of its symbols. I hacked some support for chibi but I’m not satisfies. My attempts to replace guile with chibi are here: the code start booting but the tm-defined functions are not correcly bound and there are error which prevents the interface to boot completely. Integrating s7 should not take much on the C++ side (one/two days) but still the stumbling block is how to make the scheme code compatible with the new interpreter. Best Max
|
[Prev in Thread] | Current Thread | [Next in Thread] |