Page 4 of 7

PostPosted: Tue Sep 20, 2005 12:41 pm
by renderman
If anyone has cygwin installed and wants to test, I have a WRT-Drone online, I just need the IP address you'd be coming in from to allow you access to it for testing.

Right now I juat want to make sure I have'nt forgotten any packages in the list.

PostPosted: Tue Sep 20, 2005 7:51 pm
by goldfndr
As it turns out, Cygwin's stty does seem to lie, as of Cygwin 1.5.18 and stty 5.3.0.

I verified that my setbaud code was correct by testing it on a RedHat 7.3 machine at work, then later on a Debian Sid (Kanotix) machine at home. Like stty but arguably more compact.

Now I've tried it under Cygwin, and it seems to be taking effect. "setbaud /dev/com8 38400" followed by "gpsd -n -N -D 2 /dev/com8" gets there really fast with my GPSlim236, while "setbaud /dev/com8 4800" then "gpsd -n -N -D 2 /dev/com8" takes a while longer.

So... if you don't want to wait the two minutes or so while gpsd detects your baud rate, and don't yet want to try a Cygwin snapshot, you can speed it up. Note: CYGWIN=nocom_reset might be required (that's how I have it).

Something else for y'all to test: after adding the audio libraries, compiling/installing sox, and running this device list script, I can now do "cat alert.wav > /dev/dsp" and have it generate sound - I don't know if the script or libraries or sox were really necessary, though. So, I'm anticipating that a /usr/bin/play consisting solely of
cat "$1" > /dev/dsp
will be adequate.

One more: getting flite operational:
  1. Download flite release 1.2 (source code)
  2. Download the Debian patch for flite (until I applied it, I wasn't able to successfully compile).
  3. tar vzxf flite-1.2-release.tar.gz
  4. cd flite-1.2-release
  5. patch -p1 < ../flite_1.2-release-2.diff
  6. make (there will be numerous warnings about "initialization discards qualifiers..." for cmu_us_kal_diphone.c, cmu_us_kal16_diphone.c, and one other, not sure yet how bad that is...)
  7. make install
  8. flite "hello world"


Next task: Try to figure out how to get it to lock better - gpsmodelock=true was even worse that gpsmodelock=false... and a search here for "gpsd lock" didn't help.

PostPosted: Wed Sep 21, 2005 1:41 pm
by Mark57
OK, I'm doing the test and I'm following Renderman's guide page as requested and I'm getting an error when patching gpsd.

Patch < cygwin-2.diff returns the following:

patching file configure
patching file gpsd.h
patch: **** malformed patch at line 19: others? */


I'm pretty sure I added all the Cygwin goodies, but I'll double check again.

PS, I have a few minor corrections for the guide later.

PostPosted: Wed Sep 21, 2005 1:59 pm
by renderman
Mark57 wrote:OK, I'm doing the test and I'm following Renderman's guide page as requested and I'm getting an error when patching gpsd.

Patch < cygwin-2.diff returns the following:

patching file configure
patching file gpsd.h
patch: **** malformed patch at line 19: others? */


I'm pretty sure I added all the Cygwin goodies, but I'll double check again.

PS, I have a few minor corrections for the guide later.


Make sure that when you downloaded the patch, it did'nt magle the patch file (word wrap the thing or something silly.) I caught a word wrapped comment on the file on ther server, corrected it

Minor corrections welcome

PostPosted: Wed Sep 21, 2005 2:00 pm
by Mark57
renderman wrote:Hmm, seems to be failing on serial.c. are you sure you grabbed the 2.30 source?

Also make sure that when you downloaded it, it did'nt magle the patch file (word wrap the thing or something silly.)

Minor corrections welcome


I pulled it from the link on your guide. ;) I'll pull it down again.

PostPosted: Wed Sep 21, 2005 2:01 pm
by renderman
[quote="Mark57"]I pulled it from the link on your guide. ]


I think I fixed the word wrap issue

PostPosted: Wed Sep 21, 2005 2:12 pm
by Mark57
I pulled it down again.

Now I get

Hunk #1 succeeded at 21313 (offset 65 lines)
patching file gpsd.h
patching file serial.c
patch: **** malformed patch at line 60: (speed_t)session->saved_baud);

PostPosted: Wed Sep 21, 2005 2:28 pm
by renderman
[quote="Mark57"]I pulled it down again.

Now I get

Hunk #1 succeeded at 21313 (offset 65 lines)
patching file gpsd.h
patching file serial.c
patch: **** malformed patch at line 60: (speed_t)session->saved_baud)]

Yet another word wrap. fixed. fixed a few others. try again.

Stupid vbulletin won't let me wget the diff file directly, so I have to download it with a browser and its getting wordwrapped.

PostPosted: Wed Sep 21, 2005 2:38 pm
by Mark57
renderman wrote:Yet another word wrap. fixed. fixed a few others. try again.

Stupid vbulletin won't let me wget the diff file directly, so I have to download it with a browser and its getting wordwrapped.


That fixed it! Thanks. ;) I removed the GPSD install and did it again to test from scratch. It patched just fine.

I'm trying to test this from scratch like a new user would.

My pager is going wild. Be right back.

PostPosted: Wed Sep 21, 2005 6:52 pm
by goldfndr
renderman wrote:Yet another word wrap. fixed. fixed a few others. try again.

Stupid vbulletin won't let me wget the diff file directly, so I have to download it with a browser and its getting wordwrapped.

Hmm, Firefox seemed to get the file okay, but that was on Windows... and with bugmenot accounts banned... <ducking> Knowing this can be an issue, I'll mirror any further patches on my site.

Drove home last night with gpsd, /usr/bin/play (one-line script above), and flite enabled - wow, it was pretty weird hearing sound effects... is that what you guys are used to? kismet (or, more likely, the WAV playing) segfaulted at some point during the first run, but I just ran it again and it seemed okay. flite was speaking for a while but then it got stuck on a Days Inn AP - didn't seem to have any ctrl-codes, so I don't know what the issue is.
The pronunciation seemed slightly better than Microsoft Mike but definitely worse than Crystal and Audrey, so I'll definitely find a way to adapt ns04thread.vbs for kismet. Either that or patch kismet to behave more like ns04speech.vbs. ]ls[/b] (e.g. ls /dev/com1. So, if ls /dev/dsp displays a file then some of the stuff I mentioned (e.g. compiling sox) probably isn't required.

To those of you who have gpsd running (cygwin or linux, any version), could you try cgps? I'm trying it here, and it's mostly giving me a 2D fix - I'm getting GGA followed immediately by RMC, and while the GGA does give the altitude and a 3D fix, the RMC immediately clears the altitude - this must be why Kismet isn't giving me a fix most of the time, and the only times I get a fix are when I luck out on the timing being between GGA and RMC. Need to check the code more closely, I'm guessing it's more likely a gpsd bug than a Cygwin issue.

In other news, I've successfully compiled gpsdrive within Cygwin:
  1. Downloaded packages gtk2-x11-devel, gtk2-x11, pango-devel, atk-devel, pkgconfig, gettext-devel (possibly others, possibly some of these aren't actually required)
  2. ./configure --disable-garmin
  3. make sure line 97 or so of libtool says AS=as instead of AS=, lest you get dlltool errors with a not found assembler
  4. make
  5. make install
  6. remove gpsdrive's bastardized gpsd, use this more recent copy
But it isn't finding gtk... hopefully that's easily done. Apologies if I'm leading anyone on the wrong track, but if people here have experience with X11 under Cygwin... ;)

PostPosted: Wed Sep 21, 2005 7:05 pm
by renderman
[quote="goldfndr"]
In other news, I've successfully compiled gpsdrive within Cygwin:
  1. Downloaded packages gtk2-x11-devel, gtk2-x11, pango-devel, atk-devel, pkgconfig, gettext-devel (possibly others, possibly some of these aren't actually required)
  2. ./configure --disable-garmin
  3. make sure line 97 or so of libtool says AS=as instead of AS=, lest you get dlltool errors with a not found assembler
  4. make
  5. make install
  6. remove gpsdrive's bastardized gpsd, use this more recent copy
But it isn't finding gtk... hopefully that's easily done. Apologies if I'm leading anyone on the wrong track, but if people here have experience with X11 under Cygwin... ]

Holy crap. You are amazing. I'll see about getting X11 installed to test.

PostPosted: Thu Sep 22, 2005 3:54 am
by Mark57
Long night. We lost cooling to our main data center. Had to scrounge fans, lots of them. I'm cooked. It was only 145 degrees in there. :( Good way to loose weight temporarily. ;)

I now have GPSD running with Kismet under Cygwin. Good job guys!

PostPosted: Thu Sep 22, 2005 6:44 am
by goldfndr
renderman wrote:Holy crap. You are amazing. I'll see about getting X11 installed to test.

Thanks. As an alternative, some gtk2 packages for win32 are available. But I'm not sure I'll be able to try them before next week (major upgrade at work this weekend).

Mark57 wrote:I now have GPSD running with Kismet under Cygwin. Good job guys!

Great! With that confirmation, I'm going to try submitting the patch; wish me luck. Also, I'm wondering: are you (or anyone else) able to maintain a 3D lock? cgps would be a good alternative test. I'm still wondering if it's a Cygwin issue vs gpsd issue.

PostPosted: Thu Sep 22, 2005 6:50 am
by Mark57
[quote="goldfndr"]Thanks. As an alternative, some gtk2 packages for win32 are available. But I'm not sure I'll be able to try them before next week (major upgrade at work this weekend).


Great! With that confirmation, I'm going to try submitting the patch]

I'll check when I get home. Still working on the data center at work. I'm using the USGlobalSat BU-353 which has the same internals as the Holux 236.

PostPosted: Thu Sep 22, 2005 12:12 pm
by Mark57
[quote="goldfndr"]Thanks. As an alternative, some gtk2 packages for win32 are available. But I'm not sure I'll be able to try them before next week (major upgrade at work this weekend).


Great! With that confirmation, I'm going to try submitting the patch]

No, I do not get or keep a 3D lock. Just tested it again.