Comments on: Phases of source analysis

By: Conor

Conor — Sun, 12 Jun 2005 10:35:21 +0000

That sounds like something which will connect up well to the new structure, where we communicate with tree structures which use a private source of freshness to distinguish or identify names.

By: Cyril Konopko

Cyril Konopko — Sun, 12 Jun 2005 10:26:22 +0000

By saying “Epigram response”, I mean its response on “inspect” query. Other responses my editor just doesn’t read. Of course, it’s wrong.
Syntactic correctness is guaranteed by editor, so escaping from a shed is always a success – but if term having been sent to Epigram is wrong-typed, my editor doesn’t know nothing about that – and this is another one big problem, which I’d rather not to solve, if I have ability to send and receive tree nodes instead of text.
What about scoping, it’s made using back references, hence I don’t deal with scoping by name: semantically different identifiers may share the same name wihout shadowing or overriding each other or something. If Epigram wants to have fresh names or even wants to have any presumptions about names, I’ll have to develop something more to support those presumptions. However, every identifier (every node, in fact) in my editor, has its unique internal number, which can serve the purpose to provide fresh names. That is, every occurence of an identifier is just a reference to its binding or declaration, which contains user-defined arbitrary name and machine-generated unique number.

By: Cyril Konopko

Cyril Konopko — Sun, 12 Jun 2005 08:32:37 +0000

That editor is written using a tool called MPS (Meta Programming System) which is currently developed by JetBrains, and does not run without MPS.

It will be EAP of MPS soon, then MPS will be available at Internet. When it is, I can provide you with .mpr files – by making an Internet page for them, I think.

Currently, it has one big mistake – I think, it does not parse properly Epigram response when it’s to occupy more than one line, if rendered by Emacs. I don’t know, if to fix that somehow or just to wait for convenient editor-and-Epigram interface.

By: txa

txa — Fri, 10 Jun 2005 10:48:50 +0000

Cyril, would you mind to tell us more about your editor? Is it available for playing around with it?

By: Cyril Konopko

Cyril Konopko — Thu, 09 Jun 2005 10:12:23 +0000

As far as I can understand, my editor is to connect to the elaborator directly. Scoping is processed on my editor’s side. The result of processing is a kind syntax tree with back references to variables’ bindings or declarations.

Now it interacts with Durham version of Epigram, by pretending to be XEmacs. It performs that in very silly way, in fact: it renders its result tree into text and emulates sending it to Epigram shed by Ctrl+C Ctrl+C… Hence, editor only guarantees syntactical and scoping correctness of program, but all scoping and parsing information is lost while sending result to Epigram.

That’s why I’m waiting very much for those ’services’ to be provided by Epigram1.