trb@stag.UUCP ( Todd Burkey ) (02/02/89)
[I'm adding the atari.st group to this discussion, since FOLDED will be released on the ST at the same time as the Unix version is...if and when I find time to finish it. Since it sounds like there already is a folding editor for the MAC, I'll probably tweak it for the IBM PC next. -Todd] In article <1385@raven.ukc.ac.uk> arc1@ukc.ac.uk (A.R.Curtis) writes: > >[mentions one type of folding using pointers to text files] >[and...] >GNU emacs might be a good starting point in fact, since we can use >text maps to define the fold structure (no nasty BEL characters etc) >and then OPEN FOLD/ENTER FOLD invoke functions to manipulate this >map. The map could either prefix the file (as is done in emacs rmail >and news interfaces) or be a separate file altogether (cf. the MAC >folding editor which I think uses separate files). Defining and >selecting regions would give a nice way of representing the folds. If anyone is interested, I did implement a simple minded folding mechanism for the Apollo DM editor and vi quite a while back. The approach I used was to simply allow the user to mark a region, and then invoke an external command called 'fold' which took the marked region, created a file for the region in a 'folded' directory, and returned a single line of text to the editor such that: int working_proc() { i=i+1; return(i); } becomes: %include "folded/fld001";/* int working_proc() */ By attaching a similar program to a key that unfolds the 'current' line, and by defining still a third key to toggle between three different fold types (C, Pascal, and Troff), it actually turned out to be a useful tool on the Apollos. If there is enough interest, I will dig out the code and post it. The main reason I decided I needed a folding editor rather than stay with the above methodology was simply that I wanted to be able to use the same folding mechanism on all the systems I program on (plus having a seperate file for each fold tends to get a bit slow and messy). FOLDED does create a control file for each file that has folds in it (I originally tried a single control file per directory, but that gets messy when you want to move things around.) The program also will optionally create a configuration file when you exit it such that the next time you invoke FOLDED it will open all the files you were previously editing and place you at the same locations (I used to do this with Emacs). All these extra files may be a bit messy, but they are a LOT better than trying to deal with embedded characters in your source code. I am not sure how easy a true folding capability would be to add to Emacs. To obtain efficient management of the folded text, you really need to build a linked list of all the folds (assuming you are modifying the Emacs source code...and I'm further assuming that Emacs uses a next/prev list structure for lines). If you are doing a Lisp level implementation of the folding, you may go insane trying to keep the folding process managed using a text map. I felt I was losing it many a night while trying to keep all the lists and pointers sane every time something internal got realloced or freed. Some things I still haven't decided on yet for the editor: 1) is it worth searching inside of folded text as well as out (and if something is 'found', then what is the best way to present it to the user)...maybe just a temporary 'unfoldall' mode would suffice. 2) ditto for global substitutes, deletes, etc. 3) slow terminal support...does anyone out there actually bother to even use visual editors over modems slower than 2400 baud? 4) graphics window support...I am very tempted to put in hooks to allow multiple buffers to pop up seperate windows in something like the Bellcore mgr package (X and NeWS still seem too slow and bulky to bother with). Has anyone out there programmed on top of mgr yet? -Todd Burkey "A member of STdNET-The ST developers' Network" trb@stag.UUCP
t19@np1.hep.nl (Geert J v Oldenborgh) (02/03/89)
It may be of interest to hear that a folding editor for the Atari has been for sale for the last 1.5 year or so. It implements folds as special lines that start with three arbitrary characters (to comment out the line), followed by #[ name: (beginning), #] name: (end). This leaves all folding in the editor, so the speed is OK (instanteneous for all operations). A UNIX port is being beta-tested right now, the PC port is in the same stage. As my linker is much slower than the compiler I tend to have large files and I cannot imagine how awful editing was without folds ... * #[ signature: Geert Jan van Oldenborgh t19@nikhefh.hep.nl * #] signature: