|
From: Gilboa D. <gi...@gm...> - 2007-07-31 10:52:00
|
On Mon, 2007-07-30 at 15:53 -0400, Bob Rossi wrote: > At a minimum, I at least ask each new developer if it's OK if this is > licensed under the GPL and if it's OK that I license it under a > different open source (http://www.opensource.org/licenses) approved license > in the future if I decide I want to. As long as it's OSS approved, the patches are yours. > > One other change, when you modify the behavior of CGDB, it's > important to put that information into the manual. You can modify > doc/cgdb.texinfo and then type 'make'. That will generate cgdb.txt > which you can then look over. If you have all the software necessary > you can type 'make pdf' and it will make cgdb.pdf. For a trivial > change like you have here, it should be fairly easy to do. Before I release the final patch-set, I'll create a doc/pdf/man patch as well. (Though I'm not a native English speaker, so it'll be wise to double check my grammar/syntax) Either way, as requested, I'm breaking the mega patch into a series of small(er) patches. Patch 1: Replaces the current command interface (non-sorted action_t array) with a sorted array with smarter callback functions. This should reduce the coder (and performance) over-header of introducing more/complex gdb commands into the "vi" interface. (Or should I call it the cgdb interface?) If/when this patch is approved, I'll push the third patch - which adds additional commands (and keystrokes) to the "vi" interface. Patch2: - if (ioctl (fileno (stdout), TIOCGWINSZ, &screen_size) == -1) + /* Running under cgdb/ddd fix. */ + if ((ioctl (fileno (stdout), TIOCGWINSZ, &screen_size) == -1) || + (!screen_size.ws_col) || + (!screen_size.ws_row)) > I don't really understand the above change. Why was that necessary? > What exactly does the comment mean? This is rather ugly patch. ... But if you run cgdb under a second debugger that has no active tty (E.g. Under gdb or ddd with no output window), the ioctl doesn't fail, it just returns zero-filled screen_save (which in-turn, SIGSEGV all the WIDTH-1 functions) Question is, do we need to abort if no valid tty is present? - Gilboa |
