Re: Lift --stdout restriction for using reachability bitmap in pack-objects?
On Feb 17, 2015, at 02:42, Jeff King wrote: On Tue, Feb 17, 2015 at 05:36:30PM +0700, Duy Nguyen wrote: On Tue, Feb 17, 2015 at 5:13 PM, Jeff King p...@peff.net wrote: If the only reason is for gdb, then perhaps: set args pack-objects --stdout /dev/null /dev/null in gdb would help? Right. I used gdb --args command /dev/null instead. Stupid question. Sorry for the noise. I've made the same mistake myself many times. I really wish gdb would interact over /dev/tty by default. The perl debugger does this, and I find it quite handy. But I've never managed to make gdb do it. Maybe there is an option I've missed[1]. You may want to try cgdb. It is a curses front-end to gdb. I almost never run bare gdb anymore. It has a source file pane (no more line- by-line), a gdb pane and, although a bit clumsy, a TTY pane (not shown in the screen shot) that you can interact with the debuggee with. It's been around for a while, so debian-derived distros can usually just `apt-get install cgdb`. There's a screen shot available [1]. -Kyle [1] http://cgdb.github.io/ -- To unsubscribe from this list: send the line unsubscribe git in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Lift --stdout restriction for using reachability bitmap in pack-objects?
Commit 6b8fda2 (pack-objects: use bitmaps when packing objects - 2013-12-21) turns off reachability bitmap optimization if --stdout is not specified. I wonder if we could lift this restriction? My use case is debugging pack-objects in gdb (repeatedly) where this optimization could really save time. But specifying --stdout would clutter gdb interface.. I don't know this code well enough to judge, but I think the worst thing is we don't write .idx file out (maybe because it would involve a lot more work). -- Duy -- To unsubscribe from this list: send the line unsubscribe git in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: Lift --stdout restriction for using reachability bitmap in pack-objects?
On Tue, Feb 17, 2015 at 05:36:30PM +0700, Duy Nguyen wrote: On Tue, Feb 17, 2015 at 5:13 PM, Jeff King p...@peff.net wrote: If the only reason is for gdb, then perhaps: set args pack-objects --stdout /dev/null /dev/null in gdb would help? Right. I used gdb --args command /dev/null instead. Stupid question. Sorry for the noise. I've made the same mistake myself many times. I really wish gdb would interact over /dev/tty by default. The perl debugger does this, and I find it quite handy. But I've never managed to make gdb do it. Maybe there is an option I've missed[1]. The downside of what approach, though, is that you cannot restart the program reliably from within the debugger (it cannot know how to re-set-up the descriptors). So you have to quit and restart, which loses any breakpoints, etc. -Peff [1] Having written that, I'm not sure I ever tried a script that does something like: program=$1; shift exec 30 exec 41 gdb /dev/tty /dev/tty -ex set args 3 4 gdb $* $program I think that would work, but it does screw up quoting/whitespace in your arguments. -- To unsubscribe from this list: send the line unsubscribe git in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: Lift --stdout restriction for using reachability bitmap in pack-objects?
On Tue, Feb 17, 2015 at 05:07:36PM +0700, Duy Nguyen wrote: Commit 6b8fda2 (pack-objects: use bitmaps when packing objects - 2013-12-21) turns off reachability bitmap optimization if --stdout is not specified. I wonder if we could lift this restriction? I'm not sure what else would break if we did. For instance, would bitmaps kick in when doing the Counting objects phase of git repack -ad. And if so, what would the resulting pack look like? So I'm not opposed to it in principle, but I suspect there's some cleanup work to be done. But... My use case is debugging pack-objects in gdb (repeatedly) where this optimization could really save time. But specifying --stdout would clutter gdb interface.. I don't know this code well enough to judge, but I think the worst thing is we don't write .idx file out (maybe because it would involve a lot more work). If the only reason is for gdb, then perhaps: set args pack-objects --stdout /dev/null /dev/null in gdb would help? -Peff -- To unsubscribe from this list: send the line unsubscribe git in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: Lift --stdout restriction for using reachability bitmap in pack-objects?
On Tue, Feb 17, 2015 at 5:13 PM, Jeff King p...@peff.net wrote: If the only reason is for gdb, then perhaps: set args pack-objects --stdout /dev/null /dev/null in gdb would help? Right. I used gdb --args command /dev/null instead. Stupid question. Sorry for the noise. -- Duy -- To unsubscribe from this list: send the line unsubscribe git in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html