Codestin Search App

gautierhattenberger · 2016-01-08T21:56:56Z

New external message library and tools. The main goal is to allow third party applications to communicate more easily with Paparazzi.
It currently provides:

messages definition (custom messages file can still be used)
ocaml and python libraries for ground or embedded CPU usage
C code generator (written in python, based on mavlink generators) to generate C embedded code

Regarding the integration to the main project:

all airborne code is converted
all ocaml programs are converted
audio and aerocomm telemetries have been removed (obsolete)

This pull request is currently at initial review stage and might need some more cleaning

some stuff are compiling but not working (at least sim) needs a lot of cleaning needs a better way to init transport types

fix conflicts and update bluegiga datalink event

gautierhattenberger · 2016-01-14T16:30:37Z

Pprzlink should be mostly functional now. Their is one remaining question: should we use the message definition from paparazzi (good old conf/messages.xml) or use the one provided by pprzlink (message_definitions/v1.0/messages.xml in the pprzlink tree). Both are possible
Option one might avoid people from loosing their custom messages, option two limits duplicated work, since the two file should be synced. Also custom modifications from pprzlink file might be troublesome with the perspective of binary install.
Option three: use messages.xml from paparazzi if it exists, fallback to pprzlink otherwise

All comments are welcome, since this stuff is a bit blocking for other PR like #1509

dewagter · 2016-01-14T16:58:53Z

Custom messages are a pain even with conf/messages.XML

What about creating an option to include personal messages in pprzlink and
then using only that. E.g. reserve last 10 message id's for an included
personal.XML
On Jan 14, 2016 5:30 PM, "Gautier Hattenberger" [email protected]
wrote:

Pprzlink should be mostly functional now. Their is one remaining question:
should we use the message definition from paparazzi (good old
conf/messages.xml) or use the one provided by pprzlink
(message_definitions/v1.0/messages.xml in the pprzlink tree). Both are
possible
Option one might avoid people from loosing their custom messages, option
two limits duplicated work, since the two file should be synced. Also
custom modifications from pprzlink file might be troublesome with the
perspective of binary install.
Option three: use messages.xml from paparazzi if it exists, fallback to
pprzlink otherwise

All comments are welcome, since this stuff is a bit blocking for other PR
like #1509 #1509

—
Reply to this email directly or view it on GitHub
#1513 (comment).

gautierhattenberger · 2016-01-14T17:03:50Z

this will be part of the next version of the protocol, I want to avoid this kind of modification for v1.0

dewagter · 2016-01-14T17:21:01Z

IMHO: the kindest approach would be:

step 1: be able to select which messages.xml one you want to use: default
new pprzlink
(so people with custom messages are not screwed)
step 2: a.s.a.p. add custom messages to pprzlink v2.0
step 3: remove ./conf/messages.xml

(person

-Christophe

On Thu, Jan 14, 2016 at 6:03 PM, Gautier Hattenberger <
[email protected]> wrote:

this will be part of the next version of the protocol, I want to avoid
this kind of modification for v1.0

—
Reply to this email directly or view it on GitHub
#1513 (comment).

gautierhattenberger · 2016-01-14T17:33:16Z

So how do you see step 1 without step 3 ? removing messages.xml is harmless if you don't have modifications and will keep it (with some complains) if you have some. But if you have both, how to you select ? Editing makefile is a pain, exporting env variables or using different makefile rules is error prone.

dewagter · 2016-01-14T19:09:50Z

Symlink?

Maybe make a poll to see if all steps are wanted by many. Personally I
would do the hard switch right away.
On Jan 14, 2016 6:33 PM, "Gautier Hattenberger" [email protected]
wrote:

So how do you see step 1 without step 3 ? removing messages.xml is
harmless if you don't have modifications and will keep it (with some
complains) if you have some. But if you have both, how to you select ?
Editing makefile is a pain, exporting env variables or using different
makefile rules is error prone.

—
Reply to this email directly or view it on GitHub
#1513 (comment).

gautierhattenberger · 2016-01-14T21:26:16Z

Then I vote for removing old messages.xml and use the new one unless a custom one exist.
Next step (for an other PR) would be to update pprzlink version (protocol, messages, ...)

flixr · 2016-01-14T21:28:18Z

How do you want to check if a custom one exists?
I would be happy with any solution where we don't have multiple messages.xml that can get out of sync...

gautierhattenberger · 2016-01-14T21:31:12Z

just checking if a file $(PAPARAZZI_HOME)/conf/messages.xml exists from the makefile and if yes pass it to pprzlink to overwrite default value.

and var/messages.xml for things when it's actually built

flixr · 2016-01-14T21:52:11Z

Ok, I think that should work... as long as nothing directly uses conf/messages.xml anymore...

- by default use pprzlink message definition - if custom messages are need to be used, place the file messages.xml in conf - only use default units.xml file - better quiet compilation

flixr · 2016-01-15T20:22:00Z

@gautierhattenberger there are still some warnings that XML validation is not performed.
Seems this is because the respective libs are missing in pprzlink...
Was that intentional?

gautierhattenberger · 2016-01-15T20:35:34Z

kind of
it requires some extra python lib not in the dependency of paparazzi-dev yet. Currently I'm using a parser that seems to come with standard Python stuff and that is used by mavlink. But it is maybe not the best choice and maybe we should switch to lxml that I think we already use. Also lxml has native DTD check support while the library used for mavlink is using xsd schema.

flixr · 2016-01-15T21:11:04Z

AFAIK xsd is more powerful/descriptive than dtd and it seems that mavlink includes the minixsv lib in the tree...

I quickly tried to add the libs and then it will try to validate (takes a while though), but it fails since the xsd doesn't completely match our allowed values in the messages.xml...

In any case getting those warnings is not so nice if it is not even supported by us...

flixr · 2016-01-15T21:13:24Z

See paparazzi/pprzlink#1 for validation...

gautierhattenberger · 2016-01-18T14:22:31Z

What else is missing ? Does anyone has some time to try/comment this ? @dewagter @fvantienen @esden ?

instead of disabling validation in pprzlink by default

flixr · 2016-01-23T16:17:13Z

Looks good to merge to me...

Pprzlink Here we go!

gautierhattenberger and others added 27 commits December 26, 2015 23:48

[pprlink] start using pprzlink for airborne code

cfbbf0d

some stuff are compiling but not working (at least sim) needs a lot of cleaning needs a better way to init transport types

update pprzlink module

d30f303

update pprzlink, sim with ivy is working

88999c0

update pprzlink for latest message definition

450c6d1

[pprzlink] update telemetry makefiles

107a08c

fix header include in nps

e96b4e4

update pprzlink

b8127b7

[pprzlink] clean makefile

1b21a2b

[pprzlink] update ocaml lib install

fc10fee

[pprzlink] move installed C files to var/share/pprzlink/src

d8bbf67

[pprzlink] fix pprzlog and check functions

8fb41e6

[pprzlink] add pprzlink dependency to lib_pprz and convert link agent

7b7756c

[pprzlink] fix linking with pprzlink C stub

1c383e8

[pprzlink] convert all ocaml programs to pprzlink

0cf0595

[tmtc] remove multimon from some targets

dc44bdb

[ground_segment] remove multimon (telemetry over audio)

3080a62

[pprzlink] generate intermcu messages from pprzlink

85cf59e

[pprzlink] start fixing some compilation issues

c049e4b

[pprzlink] fix transport files path

8c1694f

[pprzlink] fix path for test progs

e1a3ae3

[pprzlink] fix conpilation errors with some devices

b726c0f

make sure pprzlink submodule is initialized before building

a953998

[ext] add pprzlink convenience targets

1a935e5

[pprzlink] fix datalink include

e6ce77d

[pprzlink] more datalink include fixing

0d09106

Merge remote-tracking branch 'paparazzi/master' into pprzlink

e0258e1

fix conflicts and update bluegiga datalink event

[datalink] fix bluegiga parser

9a9f4a0

[python] remove pprz_msg and use pprzlink instead

50dcf96

[tools] gen_messages_doc.py: use messages.xml from pprzlink by default

1bdc208

and var/messages.xml for things when it's actually built

gautierhattenberger added 2 commits January 15, 2016 13:17

[pprlink] remove paparazzi message definition

dc30874

- by default use pprzlink message definition - if custom messages are need to be used, place the file messages.xml in conf - only use default units.xml file - better quiet compilation

[pprzlink] point to messages.xml file in var directory

a9c7b95

[pprzlink] update submodule

c9a9271

[pprzlink] call pymessages with VALIDATE_XML=FALSE

cd0dcfb

instead of disabling validation in pprzlink by default

gautierhattenberger added a commit that referenced this pull request Jan 23, 2016

Merge pull request #1513 from paparazzi/pprzlink

95231ed

Pprzlink Here we go!

gautierhattenberger merged commit 95231ed into master Jan 23, 2016

gautierhattenberger deleted the pprzlink branch January 23, 2016 21:10

Uh oh!

Conversation

gautierhattenberger commented Jan 8, 2016

Uh oh!

gautierhattenberger commented Jan 14, 2016

Uh oh!

dewagter commented Jan 14, 2016

Uh oh!

gautierhattenberger commented Jan 14, 2016

Uh oh!

dewagter commented Jan 14, 2016

Uh oh!

gautierhattenberger commented Jan 14, 2016

Uh oh!

dewagter commented Jan 14, 2016

Uh oh!

gautierhattenberger commented Jan 14, 2016

Uh oh!

flixr commented Jan 14, 2016

Uh oh!

gautierhattenberger commented Jan 14, 2016

Uh oh!

flixr commented Jan 14, 2016

Uh oh!

flixr commented Jan 15, 2016

Uh oh!

gautierhattenberger commented Jan 15, 2016

Uh oh!

flixr commented Jan 15, 2016

Uh oh!

flixr commented Jan 15, 2016

Uh oh!

gautierhattenberger commented Jan 18, 2016

Uh oh!

flixr commented Jan 23, 2016

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants