Re: [ssax-sxml] SXPath and GUILE
Brought to you by:
oleg
From: <da...@al...> - 2003-01-07 11:31:58
|
On Mon, Jan 06, 2003 at 10:24:34AM +0100, Joerg F. Wittenberger wrote: > Maybe we should talk about project goal and management questions > before. Askemos http://www.askemos.org is a distributed agent > programming environment where all computation is synchonized through > byzantine protocols and no super user power exists. Therefore it's > the first intrusion resistant and practical deployed operating system. Well I am not in position to talk project management on behalf of TeXmacs. Still we can see if there is room for cooperation. I had a very quick look at www.askemos.org. TeXmacs might be interested in Freenet-like networks in the future since one of the goals of the project (but actually not part of my personal goals) is to develop a cooperative publication system. If the internet become a more difficult place for free speech (thanks to DMCA, software patents and the like), peer-to-peer networks might become useful. Still I do not see how TeXmacs could benefit from a distributed computing system in the next years. > Currently we form the Askemos Network ( > http://www.softeyes.net/AskeosNetwork ) - a coalition of providers. > This implies that more power is coming behind step by step. I'd be > glad to welcome texmacs cooperation in any common interest. Looks very much like at work in progress at the moment :-) > Currently only a web browser based interface exists for Askemos and we > have to few applications. I dream up a more friendly and controlable > userinterface. Until now I mostly considered to use XUL from mozilla. > But actually I also considered before to look into texmacs. If we > could expose some access to the widget infra structure via SOAP calls Ha, here it comes. You are raising a difficult point. The whole GUI infrastructure of TeXmacs is going to undergo massive changes in the next year (driven by the needs of porting to win32). Joris (the main author) also announced his intention to unify the document tree and widget tree. Essentially that should bring someting comparable to XUL, but I have no idea how he plans to do that. Also there are plan for a closer integration with HTTP (among other thing to be able to support WebDAV). The implementation candidate at the moment is libcurl. So that is going to be a purely client-side HTTP support. I am not sure what you are talking about when you "access to the widget infra-structure via SOAP calls". Do you mean that a remote agent send SOAP messages to TeXmacs to update a document (possibly including widgets) structure? That *might* fit in the CAS-interface infrastructure, but that is a part of the app I know very little about. The idea would be to run a separate SOAP server which would communicate through pipes with TeXmacs. I have read the CAS interface protocol allows external applications to execute Scheme code inside TeXmacs, so remote control of the GUI should be essentially possible. But I do not know what are the planned evolutions for that subsystem. To make it simple. At the moment, it is not possible to do what you want with TeXmacs. Things will probably be different in one year. But I do not believe you would have the time to get into TeXmacs code to make things happen faster. > > My XSLT is a bit rusty now so I do not remember the correct > > terminology. I was referring to the fact that when a given node is > > matched by several templates in a XSL stylesheet, the applied template > > is the one which has the most 'specific' pattern. > > Ah, I understood you would refer to some guile feature, didn't you. > If so, maybe this could help for the Askemos code. Not at all. I was just talking about stylesheet semantics for static (non-incremental) transformations. I think I can go incremantally where I want by building on an extended version of pre-post-order. > I see. In that case I can recomment the SXPath code. When I > integrated it with the Askemos code, the application programmers where > pleased. SXPath is more convinient to write that DSSSL expressions. I think I am going to be happy with SXPath. I just need to stay focused on my task now. > When integrating with the SXPath code with Askemos, I had to modify > it, and those changes unfortunately did not make it back into the SXML > code: the SXML code is only almost data model agnostic, but implements > the SXML data model. Askemos inherits from the '93 SGML tranformation > tool 'sdc' (bigloo based) some ideas for a lazy evaluated data model > (can't be done with SXML as far as I can see), even though this is not > in the Askemos code for other reasons (i.e, the code currently uses > the quite favorable persistent storage feature from rscheme - one of > two points, which make the rscheme version currently *faster* than the > corresponding chicken compiled version! this storage can not store > promises ['delay'ed evaluation]). To keep the "data model agnostic" > property, the SXPath code of Askemos doesn't refer to the SXML data > model. Instead there are a few data definitions, whic currently > support 3 different data models, SXML is easy to add that way. Hum... damn... I cannot figure out what you are talking about. I must not be experienced enough. > The parts of the Askemos source, you are directly interested in would > be mechanism/tree.scm mechanism/function/*.scm > mechanism/notation/xpath.scm and some chicken/srfi-9 code not yet > distributed. I'd send that in private mail if you want me to. Thanks for the information. But for the moment I think I am going on regular SXPath/SXSLT stuff. Thanks again for taking the time to write such a detailed answer. > > Actually am not particulary interested in XSLT compatibility. I am > > just looking for good working GPL-compatible code and > > experience-proven Scheme idioms for tree transformation (in general) > > including but not limited to the XML document model. > > Here I can recommend the Askemos code. It inherits 10 year of > experience, is referential transparent, data model agnostic, may even > be made lazy and makes the project actually cheaper, when we share the > source. I should keep an eye on that project. I subscribed to askemos-general to see what is going on. I am all in favor of more cooperation with other projects, but we already have a very full agenda (users do not stop asking for new features). Also, the TeXmacs main author likes to do things in his own way, the opening up of the development process goes very slowly. But things have improved a lot in the past year. > Let's get in touch about the project goals of texmacs and common > interests of both the projects. You are welcome on tex...@gn... to discuss this. Though Joris is away for a couple of weeks so this is not the best time to start a new discussion. http://mail.gnu.org/mailman/listinfo/texmacs-dev -- David Allouche | GNU TeXmacs -- Writing is a pleasure Free software engineer | http://www.texmacs.org http://ddaa.net | http://alqua.com/tmresources da...@al... | all...@te... TeXmacs is NOT a LaTeX front-end and is unrelated to emacs. |