home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.arch:8897 comp.lang.c:12306 comp.programming:2303 comp.unix.programmer:4290
- Newsgroups: comp.arch,comp.lang.c,comp.programming,comp.unix.programmer
- Path: sparky!uunet!cis.ohio-state.edu!zaphod.mps.ohio-state.edu!menudo.uh.edu!sugar!ficc!peter
- From: peter@ferranti.com (Peter da Silva)
- Subject: Re: Summary of responses [was: What would you like from a debugger?]
- Message-ID: <id.SKBS.RIL@ferranti.com>
- Keywords: summary,debugging
- Organization: Xenix Support, FICC
- References: <bosullvn.713613694@unix1.tcd.ie>
- Date: Thu, 13 Aug 1992 15:24:31 GMT
- Lines: 63
-
- In article <bosullvn.713613694@unix1.tcd.ie> bosullvn@unix1.tcd.ie (Bryan O'Sullivan) writes:
- > I haven't decided yet what constitutes the smallest
- > debugger I can get away with, but such things as printing data
- > structures and their contents on the fly are out of the question.
-
- Embed an interpreted language in the debugger (drive it from TCL?).
-
- Then write a program to generate a routine for printing a structure based on
- the structure contents, and bind it to a data type.
-
- Stick these in a file so the debugger can find them on demand.
-
- This is sort of like "ctags": you don't have the editor grubbing through a
- bunch of files every time you want to find a tag. It's also a more useful and
- general tool than having too many smarts hardcoded into the debugger.
-
- If you drive the debugger from TCL, someone else can write the object browser
- in TK.
-
- I don't think I want the debugger command language to be C. C isn't really
- well suited to the job. Actually, I'd kind of like a Postscript type language,
- but then I'm a Forth nut.
-
- As for finding C source, I'd like a couple of related features:
-
- 1. An interactive mode, where if it can't find a file it asks
- you where it is.
-
- 2. A tags file, containing the location of functions, that you
- can build with an external tool.
-
- In fact, having an option that puts you into an editor on the tags file
- when it can't find a source file would be a good thing.
-
- A TCL autoload file could be used for this and for the structure display
- stuff, above.
-
- > 1) a programmable interface - the human debugger should be able to
- > write some debugger script for things such as step-by-step every C
- > instruction until such condition". The debugger language should
- > absolutely be programmable. Perhaps a public domain small lisp
- > interpreter -such as Emacs lisp or Xlisp- is a good starting point.
-
- That would also be good... though I think TCL is a better command language
- than Lisp, Lisp tends to be faster. The biggest problem with lisp is the
- baroque set of basic routines.
-
- > The window interface to the debugger should also be programmable.
-
- TK? WINTERP?
-
- > Ok. I'd like a complete programming environment as a debugger (say in
- > scheme or tcl or something).
-
- Another vote for a command language!
-
- Of course if you design an API for the debugger and it's clean enough this
- would become a peice of cake. DON'T forget callbacks on breakpoints!
- --
- Peter da Silva `-_-'
- $ EDIT/TECO LOVE 'U`
- %TECO-W-OLDJOKE Not war? Have you hugged your wolf today?
- Ferranti Intl. Ctls. Corp. Sugar Land, TX 77487-5012 +1 713 274 5180
-