[settings] add optional target attribute to settings #1157

flixr · 2015-04-02T12:21:09Z

If not target attribute is given, the settings are added for all targets like before.

This makes it easier to add settings files to the conf that are only valid for some targets like nps or some that you don't want when using the test progs.

Beware of: if you compile a target that unloads some settings or modules (e.g. the sim) after you uploaded to the ap, the var/aircrafts/<ac>/settings.xml file that has all usable settings combined will change ~~~and the hence the wrong ones used in GCS and settings prog~~~.

Now the calculation of the MD5sum is taking unloaded settings into account, meaning it will change and warn the user if the last compiled target has different settings.

flixr · 2015-04-13T17:55:20Z

conf/settings/estimation/body_to_imu.xml

@@ -1,6 +1,6 @@
 <!DOCTYPE settings SYSTEM "../settings.dtd">

-<settings>
+<settings target="ap|nps">


why restrict this to ap and nps?
That means you can't use it with demo and test targets...

Because it fails with sim target where imu.c is not compiled. If you want to keep it without a specific target, I suggest to fix sim then.

Right, how about something like flixr@493fde0 to add the possibility to exclude a list of targets with target="!foo|bar"

The try is incomplete, String.get may return Invalid_argument

gautierhattenberger · 2015-04-25T17:20:55Z

do we merge this one ?

flixr · 2015-04-26T10:02:24Z

It is ok to merge for me.
But we definitely need to better handle having a different settings.xml file after compiling another target.
At the very least something like a check for the number of settings?

flixr · 2015-07-25T13:35:38Z

Should we merge this befoer v5.6 regardless of the missing checks if you had unloaded settings/modules?

gautierhattenberger · 2015-07-25T13:52:45Z

It is very useful but sadly it is not possible to make just one line insensitive in the settings selection box. At least we could add a warning when unloading settings ?

I wanted to make the line unsensitive, but it doesn't seems to be possible. Either all the lines or nothing. As a fallback, the settings is unselected. It is not perfect (some stuff might be missed, you can check it back, ...) and the settings are removed at generation time anyway. Not sure it can be done better unless someone finds the way to control the list element sensitivity one by one.

flixr · 2015-07-25T14:55:38Z

I'm not so much concerned about the selection box, that seems to "only" be an inconvenience to me.
As said before I don't like the increased probability of mismatched settings with potentially even catastrophic outcome if you unknowingly change the wrong one.

What about adding another MD5sum for only the settings and sending that with ALIVE?
The we could check this when loading settings or the GCS to see if they match...

flixr · 2015-07-25T15:28:57Z

Or better, also do the filtering for used settings (unloaded modules with settings, targets in settings) when creating var/aircrafts/foo/conf/conf_aircraft.xml of which the md5sum is computed.

I think that would be the best solution, or am I missing something obvious here?

gautierhattenberger · 2015-07-27T16:08:03Z

It seems that we are currently only removing the settings that are not selected in the pprz center. I guess it should also work if we correctly unload them when the module is not valid for the target.

flixr · 2015-07-27T16:09:51Z

Yes and the settings that are directly added but don't match the target, could you do that in this branch?

this should prevent mismatch when settings are unloaded due to unsupported targets

flixr · 2015-07-27T22:43:58Z

The filtering now only works correctly from within paparazi center if you don't enable unloaded setting again.
As you said this is is partially a problem with the the list element.
IMHO the filtering should be done properly in gen_aircraft.ml so that the generated conf_aircraft.xml will be always correct and it also works if called directly from the command line....

gautierhattenberger · 2015-07-28T17:31:24Z

I don't get it. If I force the selection of an unwanted settings file (like nps.xml for ap target), it will be removed from conf_aircraft.xml and will not appear in the generated settings as well.
So conf_aircraft.xml really hold the conf used for the last build.

flixr · 2015-07-29T10:20:20Z

Ok, nevermind my previous comment.... that indeed works as expected now.

However I think it would be better to not "uncheck" the setttings file in the paparazzi center if it is unloaded.
Otherwise you suddenly have the settings file disabled if you change to a target that supports it again...

settings are still unloaded by gen_aircraft

flixr · 2015-07-30T09:53:55Z

Actually I think the unloaded should not change the conf_aircraft.xml (and hence the md5sum) unless they have settings.
If they don't have settings, there is not reason to have a different md5sum as the settings.xml will be the same.

flixr · 2015-07-30T11:02:46Z

Nevermind, seems to work as expected.

[settings] add optional target attribute to settings f not target attribute is given, the settings are added for all targets like before. This makes it easier to add settings files to the conf that are only valid for some targets like nps or some that you don't want when using the test progs. Beware of: if you compile a target that unloads some settings or modules (e.g. the sim) after you uploaded to the ap, the var/aircrafts/<ac>/settings.xml file that has all usable settings combined will change. Now the calculation of the MD5sum is taking unloaded settings into account, meaning it will change and warn the user if the last compiled target has different settings.

FlightEnabled-Admin · 2015-07-31T23:25:29Z

Hi, I'm just getting back into paparazzi. I tried to follow the mac
paparazzi install instructions but they appear to not work, is there an
up-to-date mac install setup I can follow?

On Thu, Jul 30, 2015 at 5:05 AM, Felix Ruess notifications@github.com
wrote:

Merged #1157 #1157.

—
Reply to this email directly or view it on GitHub
#1157 (comment).

flixr · 2015-08-02T19:15:06Z

@flightenabledOrg please use the mailing list for such questions and not some totally unrelated issue...
Also please try to include as much details as possible (OS version, paparazzi version, errors you got, etc). "does not appear to work" is to unspecific to give any real help

FlightEnabled-Admin · 2015-08-05T06:19:25Z

Whoops, I apologize.

@flightenabledOrg https://github.com/flightenabledOrg please use the
mailing list for such questions and not some totally unrelated issue...
Also please try to include as much details as possible (OS version,
paparazzi version, errors you got, etc). "does not appear to work" is to
unspecific to give any real help

—
Reply to this email directly or view it on GitHub
#1157 (comment).

flixr reviewed Apr 13, 2015
View reviewed changes

flixr and others added 5 commits July 25, 2015 16:32

[settings] add optional target attribute to settings

5c325f1

add targets to body_to_imu settings

d5b8e4e

[settings] possibility to exclude targets

5102c46

[settings] prevent Invalid_argument of String.get

b3fa381

flixr force-pushed the settings_target branch from ca2e45e to b3fa381 Compare July 25, 2015 14:35

gautierhattenberger added 2 commits July 27, 2015 19:57

[settings] generate MD5SUM corresponding to the actual settings conf

bc2d687

this should prevent mismatch when settings are unloaded due to unsupported targets

[settings] fix empty settings case

949d926

flixr added the Build system label Jul 29, 2015

gautierhattenberger and others added 2 commits July 29, 2015 23:48

[settings] replace automatic un-checking by a warning

071c319

settings are still unloaded by gen_aircraft

[settings] reword warning if settings were unloaded

1e70dbf

[pprz center] print unloaded modules/settings as info and color green

f4d7f11

flixr merged commit ade0ad6 into master Jul 30, 2015

flixr deleted the settings_target branch July 30, 2015 11:05

flixr added this to the v5.6 milestone Jul 30, 2015

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[settings] add optional target attribute to settings #1157

[settings] add optional target attribute to settings #1157

flixr commented Apr 2, 2015

flixr Apr 13, 2015

gautierhattenberger Apr 13, 2015

flixr Apr 13, 2015

gautierhattenberger Apr 14, 2015

gautierhattenberger commented Apr 25, 2015

flixr commented Apr 26, 2015

flixr commented Jul 25, 2015

gautierhattenberger commented Jul 25, 2015

flixr commented Jul 25, 2015

flixr commented Jul 25, 2015

gautierhattenberger commented Jul 27, 2015

flixr commented Jul 27, 2015

flixr commented Jul 27, 2015

gautierhattenberger commented Jul 28, 2015

flixr commented Jul 29, 2015

flixr commented Jul 30, 2015

flixr commented Jul 30, 2015

FlightEnabled-Admin commented Jul 31, 2015

flixr commented Aug 2, 2015

FlightEnabled-Admin commented Aug 5, 2015

[settings] add optional target attribute to settings #1157

[settings] add optional target attribute to settings #1157

Conversation

flixr commented Apr 2, 2015

flixr Apr 13, 2015

Choose a reason for hiding this comment

gautierhattenberger Apr 13, 2015

Choose a reason for hiding this comment

flixr Apr 13, 2015

Choose a reason for hiding this comment

gautierhattenberger Apr 14, 2015

Choose a reason for hiding this comment

gautierhattenberger commented Apr 25, 2015

flixr commented Apr 26, 2015

flixr commented Jul 25, 2015

gautierhattenberger commented Jul 25, 2015

flixr commented Jul 25, 2015

flixr commented Jul 25, 2015

gautierhattenberger commented Jul 27, 2015

flixr commented Jul 27, 2015

flixr commented Jul 27, 2015

gautierhattenberger commented Jul 28, 2015

flixr commented Jul 29, 2015

flixr commented Jul 30, 2015

flixr commented Jul 30, 2015

FlightEnabled-Admin commented Jul 31, 2015

flixr commented Aug 2, 2015

FlightEnabled-Admin commented Aug 5, 2015