chiark / gitweb /
Ian Jackson [Sun, 9 Jan 2011 18:03:47 +0000 (18:03 +0000)]
gui: add X2 to keys, since otherwise you can sometimes get MovFeatTooLate annoyingly
Ian Jackson [Sun, 9 Jan 2011 17:44:53 +0000 (17:44 +0000)]
realtime: movpos: need to cope with reentrant calls to method_change_done, since FSQ does it
Ian Jackson [Sun, 9 Jan 2011 17:35:30 +0000 (17:35 +0000)]
hostside: clean removes autoversion.c
Ian Jackson [Sun, 9 Jan 2011 17:31:20 +0000 (17:31 +0000)]
hostside/x.gdb: back to debugging realtime
Ian Jackson [Sun, 9 Jan 2011 17:12:15 +0000 (17:12 +0000)]
hostside/TODO: update with some info from discoveries etc.
Ian Jackson [Sun, 9 Jan 2011 17:01:43 +0000 (17:01 +0000)]
hostside: multiplex: make realtime_xopts a command line setting
Ian Jackson [Sun, 9 Jan 2011 16:54:39 +0000 (16:54 +0000)]
hostside: multiplex: spot when restart is looping and abandon it
Ian Jackson [Sun, 9 Jan 2011 14:02:41 +0000 (14:02 +0000)]
hostside: include autoversion and print it in a couple of relevant places
Ian Jackson [Sun, 9 Jan 2011 14:01:47 +0000 (14:01 +0000)]
git-revid: new C autoversion feature
Ian Jackson [Sun, 9 Jan 2011 13:36:19 +0000 (13:36 +0000)]
realtime: persist: trivial whitespace fix
Ian Jackson [Sun, 9 Jan 2011 13:35:17 +0000 (13:35 +0000)]
hostside/TODO: some things fixed
Ian Jackson [Sun, 9 Jan 2011 13:31:47 +0000 (13:31 +0000)]
realtime: do not crash and/or recurse if die called too early
Ian Jackson [Sun, 9 Jan 2011 13:31:26 +0000 (13:31 +0000)]
hostside/x.gdb: debugging persist conversion
Ian Jackson [Sun, 9 Jan 2011 13:31:01 +0000 (13:31 +0000)]
realtime: persist: header check failures tell us which item did not match
Ian Jackson [Sun, 9 Jan 2011 13:30:29 +0000 (13:30 +0000)]
hostside: TODOs from repeated crashes last night
Ian Jackson [Sun, 9 Jan 2011 13:18:54 +0000 (13:18 +0000)]
realtime: in entrails conversion, do not try to read state of moving features out of non-saved memory
Ian Jackson [Sun, 9 Jan 2011 04:20:31 +0000 (04:20 +0000)]
realtime: movpos: avoid a use-after-free
Ian Jackson [Sun, 9 Jan 2011 04:20:13 +0000 (04:20 +0000)]
realtime: movpos: when calling movpos_change on a no-change indep, immediately finish the motion
Ian Jackson [Sun, 9 Jan 2011 03:57:28 +0000 (03:57 +0000)]
realtime: motions_all_abandon sets ->motion=0
Ian Jackson [Sun, 9 Jan 2011 02:51:11 +0000 (02:51 +0000)]
realtime: trivial whitespace fix
Ian Jackson [Sun, 9 Jan 2011 02:50:58 +0000 (02:50 +0000)]
realtime: resolve: at end of resolution, set all features we haven't explicitly changed to unknown movposcomb
Ian Jackson [Sun, 9 Jan 2011 02:26:26 +0000 (02:26 +0000)]
realtime: only monitor rusage if RTFEAT_RUSAGE
Ian Jackson [Sun, 9 Jan 2011 02:26:15 +0000 (02:26 +0000)]
realtime: rusage: print a message if we are monitoring
Ian Jackson [Sun, 9 Jan 2011 02:18:01 +0000 (02:18 +0000)]
realtime: new check_rusage feature
Ian Jackson [Sun, 9 Jan 2011 02:17:37 +0000 (02:17 +0000)]
realtime: movpos: properly terminate methods[] array (!)
Ian Jackson [Sun, 9 Jan 2011 01:52:14 +0000 (01:52 +0000)]
realtime: movpos: do not SEGV if we dispose a half-built indep
Ian Jackson [Sun, 9 Jan 2011 01:52:03 +0000 (01:52 +0000)]
realtime: movpos: debug: recategorise exit message from indep_prepare
Ian Jackson [Sun, 9 Jan 2011 01:30:55 +0000 (01:30 +0000)]
hostside: movpos: fix fundamental logic error in change_needed (!)
Ian Jackson [Sun, 9 Jan 2011 01:30:35 +0000 (01:30 +0000)]
hostside/x.gdb: more point debug
Ian Jackson [Sun, 9 Jan 2011 01:30:16 +0000 (01:30 +0000)]
realtime: movpos: debug: print some debugging in movpos_findcomb_bysegs
Ian Jackson [Sun, 9 Jan 2011 01:11:04 +0000 (01:11 +0000)]
hostside: fetch-dump: new convenience script
Ian Jackson [Sun, 9 Jan 2011 00:35:52 +0000 (00:35 +0000)]
realtime: movpos: movpos_change actually sets seg->moving (!)
Ian Jackson [Sun, 9 Jan 2011 00:25:12 +0000 (00:25 +0000)]
realtime: movpos: when movpos_reserve actually returns the reservation (!)
Ian Jackson [Sun, 9 Jan 2011 00:24:47 +0000 (00:24 +0000)]
realtime: movpos: debug output: exclude some more stuff from the default movpos output
Ian Jackson [Sun, 9 Jan 2011 00:23:33 +0000 (00:23 +0000)]
realtime: safety: debug output: properly break up the big line of segments, when we call out to movpos
Ian Jackson [Sat, 8 Jan 2011 23:45:28 +0000 (23:45 +0000)]
realtime: allow suppression of use of the PICs watchdog feature
Ian Jackson [Sat, 8 Jan 2011 23:45:14 +0000 (23:45 +0000)]
realtime: if persist_fn is empty, do not attempt persist conversion etc. (useful for valgrind, which does strange things to /proc/self/exe)
Ian Jackson [Sat, 8 Jan 2011 23:21:11 +0000 (23:21 +0000)]
realtime/movpos: fix some partially unitialised structs, and fix some non-newline-terminated debugging messages in error paths
Ian Jackson [Sat, 8 Jan 2011 23:20:36 +0000 (23:20 +0000)]
realtime/movpos: improve a debugging message
Ian Jackson [Sat, 8 Jan 2011 23:20:09 +0000 (23:20 +0000)]
hostside/x.gdb: debugging new multi-movfeat movpos.c
Ian Jackson [Sat, 8 Jan 2011 23:19:51 +0000 (23:19 +0000)]
realtime: new -B option for turning off nonblockingness of stdin, helpful for valgrind
Ian Jackson [Fri, 31 Dec 2010 13:08:44 +0000 (13:08 +0000)]
movpos: wip multiple movfeatkinds: compiles, not tested
Ian Jackson [Thu, 30 Dec 2010 14:39:06 +0000 (14:39 +0000)]
movpos: wip multiple movfeatkinds: debugging tidyup, before compile
Ian Jackson [Thu, 30 Dec 2010 13:15:28 +0000 (13:15 +0000)]
movpos: wip multiple movfeatkinds: tidying up, before add lots of DPRINTFs
Ian Jackson [Wed, 29 Dec 2010 17:04:41 +0000 (17:04 +0000)]
movpos: wip multiple movfeatkinds: new indep/meth protocol, is good. wip
Ian Jackson [Tue, 28 Dec 2010 21:48:51 +0000 (21:48 +0000)]
movpos: wip multiple movfeatkinds in one request; before abolish indep_prepare
Ian Jackson [Tue, 28 Dec 2010 01:16:58 +0000 (01:16 +0000)]
movpos: wip multiple movfeatkinds in one request; working on indep code
Ian Jackson [Sun, 19 Dec 2010 18:04:46 +0000 (18:04 +0000)]
movpos: wip multiple different movfeatkinds in one request
Ian Jackson [Wed, 15 Dec 2010 21:19:38 +0000 (21:19 +0000)]
hostside/movpos: fix some comments in waggle section
Ian Jackson [Sat, 8 Jan 2011 21:42:05 +0000 (21:42 +0000)]
gui: get list of trains available from train_direction (ie "train * at") not train_commanded ("train * speed commanding") since the latter is sent by realtime even for trains which aren't on the layout
Ian Jackson [Sat, 8 Jan 2011 21:33:01 +0000 (21:33 +0000)]
gui: show unknown movpos with black-on-white labels
Ian Jackson [Sat, 8 Jan 2011 21:32:47 +0000 (21:32 +0000)]
gui: set all movpos to unknown in state Resolving
Ian Jackson [Sat, 8 Jan 2011 21:32:26 +0000 (21:32 +0000)]
gui: make movpos-button-sstate take a posn arg and make it call -setdisplay; no functional change
Ian Jackson [Sat, 8 Jan 2011 21:29:48 +0000 (21:29 +0000)]
Revert "gui: keypress for position-unknown movfeat normalises rather than reverses"
Actually, this change did the opposite of what was intended.
This reverts commit
a2e566e03eb861afc530369ae9cfc9e38a882771.
Ian Jackson [Sat, 8 Jan 2011 21:11:32 +0000 (21:11 +0000)]
hostside/multiplex: clear movpos states from replay list on Resolving
Ian Jackson [Sat, 8 Jan 2011 20:50:13 +0000 (20:50 +0000)]
realtime: tune the retransmission table, and change the time unit to be NMRADATA bytes rather than queue entries
Ian Jackson [Sat, 8 Jan 2011 20:37:45 +0000 (20:37 +0000)]
realtime: actually do initial queueing of speedmanager's retransmit nodes (!)
Ian Jackson [Sat, 8 Jan 2011 20:37:03 +0000 (20:37 +0000)]
realtime: do not retransmit NMRA idle packets for nonexistent adjucts - instead, transmit nothing at all
Ian Jackson [Sat, 8 Jan 2011 20:36:28 +0000 (20:36 +0000)]
realtime: fix a comment which referred to "features" rather than "adjuncts"
Ian Jackson [Sat, 8 Jan 2011 20:18:04 +0000 (20:18 +0000)]
realtime: add debugging output for retransmit.c
Ian Jackson [Sat, 8 Jan 2011 19:21:58 +0000 (19:21 +0000)]
realtime: Fix typo in comment
Ian Jackson [Sat, 8 Jan 2011 19:14:55 +0000 (19:14 +0000)]
detpic/display-nmra: new NmraAssist::HighPassFilter for more reliable capture of data from track
Ian Jackson [Fri, 7 Jan 2011 20:39:54 +0000 (20:39 +0000)]
realtime: actually ever regard anything as interfering (!) (confirmed fixed)
Ian Jackson [Fri, 7 Jan 2011 20:36:56 +0000 (20:36 +0000)]
realtime: actually ever regard anything as interfering (!)
Ian Jackson [Fri, 7 Jan 2011 01:51:10 +0000 (01:51 +0000)]
hostside/gui-plan: do not crash on the first "movepos Z0 position ? moving" (untested)
Ian Jackson [Thu, 6 Jan 2011 02:05:27 +0000 (02:05 +0000)]
hostside: rename nmra functions etc. from "features" to "adjuncts" (name chosen from NAMES-FOR-FEATURE)
Ian Jackson [Wed, 5 Jan 2011 00:47:02 +0000 (00:47 +0000)]
Makefiles: remove +persist.* and +realtime.log on clean, in hostside
Ian Jackson [Wed, 5 Jan 2011 00:46:44 +0000 (00:46 +0000)]
Makefiles: run realclean as tidy first then clean, as clean likes to delete some included makefiles which recursive tidy will put back
Ian Jackson [Wed, 5 Jan 2011 00:46:09 +0000 (00:46 +0000)]
Makefiles: add some more missing cleans: *.tmp, *.new, t u v, .*~, etc., in assorted directories
Ian Jackson [Wed, 5 Jan 2011 00:38:01 +0000 (00:38 +0000)]
Makefiles: add some missing cleans
Ian Jackson [Wed, 5 Jan 2011 00:37:41 +0000 (00:37 +0000)]
Makefiles: add tidy and realclean targets to all pics subdirs; detpic removes t.*.crash on tidy
Ian Jackson [Wed, 5 Jan 2011 00:18:24 +0000 (00:18 +0000)]
gui: keypress for position-unknown movfeat normalises rather than reverses
Ian Jackson [Wed, 5 Jan 2011 00:16:44 +0000 (00:16 +0000)]
Makefiles: do not run "tidy" target by default; new realclean target does both
Ian Jackson [Wed, 5 Jan 2011 00:12:50 +0000 (00:12 +0000)]
realtime: fix recently-introduced bug where wrong sense of ec is checked
Ian Jackson [Tue, 4 Jan 2011 02:33:48 +0000 (02:33 +0000)]
hostside/gui: adjustment of point positions, more
Ian Jackson [Tue, 4 Jan 2011 02:28:38 +0000 (02:28 +0000)]
gui: listen to "movpos * position" as those are in terms of abstract features, which is what we get from layout/
Ian Jackson [Tue, 4 Jan 2011 01:57:31 +0000 (01:57 +0000)]
hostside/gui: adjustment of point positions: make many "normal" positions line up with track
Ian Jackson [Tue, 4 Jan 2011 01:50:54 +0000 (01:50 +0000)]
hostside/gui: adjustment of point positions
Ian Jackson [Tue, 4 Jan 2011 01:31:31 +0000 (01:31 +0000)]
hostside/gui: allow positioning of 0 and 1 posns of point button separately, and set many settings
Ian Jackson [Tue, 4 Jan 2011 01:12:23 +0000 (01:12 +0000)]
hostside/gui: whitespace changes and emacs mode setting in gui-layout.config
Ian Jackson [Tue, 4 Jan 2011 00:36:00 +0000 (00:36 +0000)]
hostside/TODO: two bugs known fixed
Ian Jackson [Tue, 4 Jan 2011 00:35:00 +0000 (00:35 +0000)]
hostside/TODO: found a gui bug
Ian Jackson [Tue, 4 Jan 2011 00:28:31 +0000 (00:28 +0000)]
hostside/gui: include keys for C5/B7 crossover
Ian Jackson [Mon, 3 Jan 2011 23:23:41 +0000 (23:23 +0000)]
realtime: avoid needless signalling problems related to interference
When we are considering an interfering segment, we want to first see
whether the segment has anything in it and only if so consider its
position. That avoids is calling it a problem that the interfering
segment's position is not known (eg, because it has never been set).
To support this, split segment_interferes into two functions:
segment_interferer: simply returns the possibly-interfering segment
so that we can check whether it's relevant
segment_interferer_does: does the hard work, dealing appropriately
with errors
At all but the one call site of interest we can assert that the
getmovpos callback never fails, so we provide a helper function
segment_interferes_simple
which does what segment_interferes used to (except that getmovpos
returning an error now causes an abort).
Ian Jackson [Mon, 3 Jan 2011 23:23:33 +0000 (23:23 +0000)]
realtime: do not panic if we consider an unknown movposcomb - just predict a problem
Ian Jackson [Mon, 3 Jan 2011 22:21:43 +0000 (22:21 +0000)]
hostside: add bavarian to x.gdb args
Ian Jackson [Mon, 3 Jan 2011 22:17:59 +0000 (22:17 +0000)]
hostside: add bavarian to multiplex; better homes
Ian Jackson [Mon, 3 Jan 2011 22:13:52 +0000 (22:13 +0000)]
firmware: if we fire a point just before being turned off, still note that the point was fired, send 20, etc.
Ian Jackson [Mon, 3 Jan 2011 22:13:26 +0000 (22:13 +0000)]
firmware: properly document complicated combined firing/off point states
Ian Jackson [Mon, 3 Jan 2011 21:06:25 +0000 (21:06 +0000)]
Merge branch 'master' of login.chiark.greenend.org.uk:things/Trains/trains
Ian Jackson [Mon, 3 Jan 2011 21:03:52 +0000 (21:03 +0000)]
hostside/bavarian: speed measurements: combine six separate runs of measure-speeds into one
rune was: ./measure-speeds /dev/railway AX speed 4 55 1 `seq 43 126` | tee bavarian-1reverse6.speeds etc.
Ian Jackson [Mon, 3 Jan 2011 21:02:34 +0000 (21:02 +0000)]
hostside/bavarian: speed measurements: edit out a bogus measurement (having seen the train do the wrong thing)
Ian Jackson [Mon, 3 Jan 2011 21:01:42 +0000 (21:01 +0000)]
hostside/bavarian: speed measurements, backwards
Ian Jackson [Mon, 3 Jan 2011 19:58:20 +0000 (19:58 +0000)]
hostside/measure-speeds: when positioning for slow, when we have just done slow speed, look for X7 too
Ian Jackson [Mon, 3 Jan 2011 19:32:44 +0000 (19:32 +0000)]
hostside/measure-speeds: when positioning for slow, when we have just done slow speed, do not bother with forward movement towards X8 et al
Ian Jackson [Mon, 3 Jan 2011 19:23:46 +0000 (19:23 +0000)]
hostside/bavarian: Add bavarian to TRAINS
Ian Jackson [Mon, 3 Jan 2011 19:23:36 +0000 (19:23 +0000)]
hostside/measure-speeds: when positioning for slow, check for entry to X8 too
Ian Jackson [Mon, 3 Jan 2011 19:23:11 +0000 (19:23 +0000)]
hostside/analyse-speeds: decreasing speed is only a warning
Ian Jackson [Mon, 3 Jan 2011 19:22:56 +0000 (19:22 +0000)]
hostside/bavarian: remove bogus speed measurements just made
Ian Jackson [Mon, 3 Jan 2011 19:03:18 +0000 (19:03 +0000)]
hostside/measure-speeds: repeat OFF command (can help if there was nmra buffered, eg)