Re: GNU Emacs position-bytes?

Mats Lidell writes: That's a horrible API. Boy, am I glad I don't work on that code! *Both* godef *and* GNU Emacs! In any case, `position-bytes' cannot work in XEmacs (ever!) because XEmacs uses an internal encoding that is used by nothing else in the world. This is true in principle for Emacs (its internal encoding is not Unicode compatible), but it should be good enough almost all the time in the case that a language defines its input as UTF-8, since Emacs's internal encoding is UTF-8. We already know how to do that. It has to be calculated, of course. (defun xemacs-call-godef (buffer bytepos) "Implements the subprocess stuff to call godef program. Stream BUFFER with 'binary encoding to godef." ...) (defun xemacs-godef (&optional point-before-identifier) "Call this function in the source buffer." (let ((pbi (or point-before-identifier (point))) ((work (clone-buffer (current-buffer)))) (pbi-bytes (encode-coding-region (point-min) pbi 'utf-8 work))) (encode-coding-region pbi-bytes (point-max) 'utf-8 work) (xemacs-call-godef work pbi-bytes))) is the basic idea. Possibly it would be a better idea to keep the work buffer around for multiple queries, but that depends on how frequent the queries are compared to editing the source buffer. Trying to keep the two synchronized would be horrible, so if the source buffer is edited you pretty much have to redo from scratch. I suppose we could also implement like this: (defun xemacs-call-godef (bytepos &optional buffer) "Implements the subprocess stuff to call godef program. Stream BUFFER with 'utf-8 encoding to godef." ...) (defun xemacs-godef (&optional point-before-identifier) "Call this function in the source buffer." (let ((pbi (or point-before-identifier (point))) ((work (clone-buffer (current-buffer)))) (pbi-bytes (encode-coding-region (point-min) pbi 'utf-8 work))) (xemacs-call-godef pbi-bytes))) It's not clear how much work this saves. The first version encodes the whole source buffer to the work buffer using 'utf-8, then encodes the work buffer to the subprocess using 'binary. The second encodes the first part of the source buffer to the work buffer using 'utf-8, then does it again using 'utf-8 to encode the whole buffer to the subprocess. UTF-8 encoding is slightly more expensive than binary encoding. If we implement fixed width buffers (see PEP 393[1] for Python's implementation), then encoding binary will be a no-op, and UTF-8 encoding will be much more expensive. Steve Footnotes: [1] http://www.python.org/dev/peps/pep-0393/ _______________________________________________ XEmacs-Beta mailing list XEmacs-Beta(a)xemacs.org http://lists.xemacs.org/mailman/listinfo/xemacs-beta