From nikodemus at random-state.net Sun Mar 14 10:01:42 2010 From: nikodemus at random-state.net (Nikodemus Siivola) Date: Sun, 14 Mar 2010 12:01:42 +0200 Subject: [phemlock-devel] phemlock is dead, long live hemlock Message-ID: <633d72b1003140301w41b29fc9y98fd7c355a0e157e@mail.gmail.com> So, David suggested we co-opt this list for the purposes of: http://gitorious.org/hemlock/hemlock (link for the benefit of those on the list not aware of the it) Cheers, -- Nikodemus From nikodemus at random-state.net Sun Mar 14 10:40:07 2010 From: nikodemus at random-state.net (Nikodemus Siivola) Date: Sun, 14 Mar 2010 12:40:07 +0200 Subject: [phemlock-devel] Hemlock @ Launchpad Message-ID: <633d72b1003140340t1f9c0c0es8ce89e6789f0713a@mail.gmail.com> https://bugs.launchpad.net/hemlock Cheers, -- Nikodemus From nikodemus at random-state.net Sun Mar 14 13:56:44 2010 From: nikodemus at random-state.net (Nikodemus Siivola) Date: Sun, 14 Mar 2010 15:56:44 +0200 Subject: [phemlock-devel] hemlock clone on gitorious In-Reply-To: <20100101144517.GA19871@radon> References: <633d72b0912220051y27b1ca61xead7538360f16383@mail.gmail.com> <20091227192245.GA19593@radon> <633d72b0912280103r4baa949erbeae1e073ff456a4@mail.gmail.com> <20100101144517.GA19871@radon> Message-ID: <633d72b1003140656we3eab01o567ad6d2ceb30c0f@mail.gmail.com> On 1 January 2010 16:45, David Lichteblau wrote: > ?- Although I have merged bbea34ea including the "Pull out and export > ? ?HEMLOCK:GOTO-BUFFER-START" change, I'm inclined to propose a > ? ?different plan: > > ? ?Hemlock commands are already callable as ordinary Lisp functions, > ? ?i.e. ?the function beginning-of-buffer-command already exists, and > ? ?I think GNU Emacs got it right when it made the distinction between > ? ?commands and functions as small as possible: Commands _should_ be > ? ?called in other Lisp code whenever it makes sense ?-- or to describe > ? ?it in the opposite way: It's okay to take Lisp API functions and > ? ?turn them into commands. > > ? ?So the steps are: > ? ? ?1. change defcommand so that it doesn't append -COMMAND to the > ? ? ? ? function name. > ? ? ?2. As a convention, write &optional prefix for such functions, so > ? ? ? ? callers don't have to supply a prefix where none is needed. > ? ? ? ? (And keep your change to declare it ignorable automatically.) > ? ? ?3. (optionally) change the name of commands as stated in the code > ? ? ? ? from strings to symbols. ?DEFCOMMAND would still define the > ? ? ? ? space-separated command name in the command table, so this > ? ? ? ? change would not be end-user-visible. ?But C-c C-w C-c would > ? ? ? ? work no the command name, etc. > > ? ?As a result, the function would be called beginning-of-buffer > ? ?(rather than goto-buffer-start), just like the command is already > ? ?called. ?GNU Emacs has the same name, which is a plus. I made some steps in this direction. http://gitorious.org/~nikodemus/hemlock/nikodemus-hemlock/commits/defcommand-hax Basically: 1. Existing DEFCOMMAND forms work just like they used to, and append -COMMAND to names to avoid need for wholesale renaming. 2. (defcommand ("My Foo" my-foo) ...) uses MY-FOO as the function name. 3. DEFCOMMAND also accepts empty lambda-lists, taking them to mean (&OPTIONAL #:PREFIX). IMO this leads to easier to read command when they don't use prefix arguments. I also went over some exiting commands and made them use the new-style naming: FORWARD-CHARACTER, BEGINNING-OF-BUFFER, etc. Let me know how this looks, -- Nikodemus From david at lichteblau.com Sun Mar 14 14:27:49 2010 From: david at lichteblau.com (David Lichteblau) Date: Sun, 14 Mar 2010 15:27:49 +0100 Subject: [phemlock-devel] hemlock clone on gitorious In-Reply-To: <633d72b1003140656we3eab01o567ad6d2ceb30c0f@mail.gmail.com> References: <633d72b0912220051y27b1ca61xead7538360f16383@mail.gmail.com> <20091227192245.GA19593@radon> <633d72b0912280103r4baa949erbeae1e073ff456a4@mail.gmail.com> <20100101144517.GA19871@radon> <633d72b1003140656we3eab01o567ad6d2ceb30c0f@mail.gmail.com> Message-ID: <20100314142749.GA4152@radon> Quoting Nikodemus Siivola (nikodemus at random-state.net): > I made some steps in this direction. > > http://gitorious.org/~nikodemus/hemlock/nikodemus-hemlock/commits/defcommand-hax Thanks for working on this. > Basically: > > 1. Existing DEFCOMMAND forms work just like they used to, and append > -COMMAND to names to avoid need for wholesale renaming. > > 2. (defcommand ("My Foo" my-foo) ...) uses MY-FOO as the function name. That's a good transition strategy at least. As long as we aren't switching over completely, we don't benefit from M-. on command names as written, but I guess that's okay for now. One thing to keep in mind though is that sometimes we need to find command function names from the command. The two function I can find which currently rely on the -command suffix are: - edit-command-definition M-. replacement for command names and - find-command-for-prepl-normally which makes it possible to run commands like CL-USER> ,start-slave-process or CL-USER> ,dired "/etc" in the repl. How would be fix them on the branch? > 3. DEFCOMMAND also accepts empty lambda-lists, taking them to mean > (&OPTIONAL #:PREFIX). IMO this leads to easier to read command when > they don't use prefix arguments. > > I also went over some exiting commands and made them use the new-style > naming: FORWARD-CHARACTER, BEGINNING-OF-BUFFER, etc. Nice. d. From nikodemus at random-state.net Sun Mar 14 15:21:46 2010 From: nikodemus at random-state.net (Nikodemus Siivola) Date: Sun, 14 Mar 2010 17:21:46 +0200 Subject: [phemlock-devel] hemlock clone on gitorious In-Reply-To: <20100101144517.GA19871@radon> References: <633d72b0912220051y27b1ca61xead7538360f16383@mail.gmail.com> <20091227192245.GA19593@radon> <633d72b0912280103r4baa949erbeae1e073ff456a4@mail.gmail.com> <20100101144517.GA19871@radon> Message-ID: <633d72b1003140821m62add414j92ccf73b1c143c20@mail.gmail.com> On 1 January 2010 16:45, David Lichteblau wrote > ?- from the tty-2 branch, I have merged the first commit: > > ? ? 4c000ea6 make arrow keys and pageup/down work in the tty backend > > ? ?I would like to merge the following commit, but I couldn't find the osicat > ? ?changes required: > ? ? 346d8fa4 don't allow C-y to generate a SIGTSTP That's commit 451bf6fa993ae74eb56aff80d40050111d7f19f4 Author: Nikodemus Siivola Date: Tue Dec 22 18:49:19 2009 +0200 add VDSUSP to termios constants which is in the Osicat main repo -- so I've pushed the SIGSTP fix to Hemlock. > ? ?However, I have not yet merged the other two: 57c160ff (distinct Backspace > ? ?and C-h in terminal) and 08f68169 (get rid of the arraw-key kludge). > > ? ?The reason is that I'd like to work on the Backspace/Delete some > ? ?more. ?(I'd like to keep keybindings so that C-h is equivalent to > ? ?backspace -- by default anyway. ?But I agree that distinct physical > ? ?keys shouldn't be conflated on the lowest level of key event > ? ?processing already.) Attached a cleaned-up patch that combines the essential parts of those commits. With this patch TTY matches more closely with the other backends: even without this in CLX Hemlock C-h and Backspace are distinct -- but our keybindings are (mostly, at least) set up to treat them as identical. Cheers, -- Nikodemus -------------- next part -------------- A non-text attachment was scrubbed... Name: 0001-improved-terminal-key-handling.patch Type: application/octet-stream Size: 8511 bytes Desc: not available URL: