]> rtime.felk.cvut.cz Git - orte.git/log
orte.git
8 months agoAdd shell.nix master
Michal Sojka [Mon, 31 Jul 2023 10:29:30 +0000 (12:29 +0200)]
Add shell.nix

This is a simple development environment based on nix-shell, which
contains the dependencies needed for building the project with meson.

5 years agoMerge branch 'shape_android_features' of https://github.com/Vajnar/orte
Michal Sojka [Mon, 12 Nov 2018 11:20:24 +0000 (12:20 +0100)]
Merge branch 'shape_android_features' of https://github.com/Vajnar/orte

5 years agoCompile shapes demo with meson
Michal Sojka [Mon, 12 Nov 2018 11:08:36 +0000 (12:08 +0100)]
Compile shapes demo with meson

5 years agoMerge branch 'shape_qt_fixes' of https://github.com/Vajnar/orte
Michal Sojka [Mon, 12 Nov 2018 09:53:29 +0000 (10:53 +0100)]
Merge branch 'shape_qt_fixes' of https://github.com/Vajnar/orte

5 years agomeson build: Properly handle generation of orte_config.h
Michal Sojka [Mon, 12 Nov 2018 09:10:31 +0000 (10:10 +0100)]
meson build: Properly handle generation of orte_config.h

By properly we mean copying the existing template for now. Later, we may
generate the config by letting meson detect the needed features and
generate the whole config automatically.

5 years agoTest meson build with Travis CI
Michal Sojka [Mon, 15 Oct 2018 04:59:53 +0000 (06:59 +0200)]
Test meson build with Travis CI

5 years agoAndroid Shape: Fix possible memory leaks
Martin Vajnar [Fri, 9 Nov 2018 19:20:30 +0000 (20:20 +0100)]
Android Shape: Fix possible memory leaks

Fixe as per https://developer.android.com/reference/android/content/ContextWrapper#getApplicationContext()

5 years agoAndroid Shape: Add support for multitouch gestures
Martin Vajnar [Fri, 9 Nov 2018 19:15:42 +0000 (20:15 +0100)]
Android Shape: Add support for multitouch gestures

It is now possible to move multiple shapes at the same time.

5 years agoAndroid Shape: Add locks for synchronization of the receive callback and GUI draw...
Martin Vajnar [Fri, 9 Nov 2018 19:13:42 +0000 (20:13 +0100)]
Android Shape: Add locks for synchronization of the receive callback and GUI draw threads

This commit adds the missing locks and changes setting of bound through
getBounds() object to setBounds() function as recommended by the
documentation (see: https://developer.android.com/reference/android/graphics/drawable/Drawable.html#getBounds())

5 years agoAdd Meson buildsystem support
Michal Sojka [Sun, 14 Oct 2018 21:50:10 +0000 (23:50 +0200)]
Add Meson buildsystem support

Tested only on Linux. Build works, but many things are still missing. For
example: shape demo, config options, support for other platforms, etc.

5 years agotravis: Install uncrustify via a package
Michal Sojka [Mon, 10 Sep 2018 21:41:03 +0000 (23:41 +0200)]
travis: Install uncrustify via a package

5 years agoSHAPE: Fix undefined behavior resulting in strange lock-ups
Martin Vajnar [Sun, 6 May 2018 10:49:46 +0000 (12:49 +0200)]
SHAPE: Fix undefined behavior resulting in strange lock-ups

This commit fixes intermittent freezes of 'subscriber' dialog. These
freezes were caused by invoking QWidget::update() directly from a non-GUI
thread. This is undefined behavior and leads to many strange lock-ups and
even SEGFAULTS.

In order to fix this we now introduce the signal changed(), that will be
emitted if repaint of shapes is requested either from publisher or
subscriber. This signal is connected to the update() slot. By default
signals are connected to slots by Qt::AutoConnection connection type. This
should ensure, that QT decides internally if it should treat this
connection as Qt::DirectConnection in case of publisher and
Qt::QueuedConnection in case of subscriber. This decision is made when the
signal is emitted.

8 years agoRTEMS: disable use of CLOCK_MONOTONIC for RTEMS.
Pavel Pisa [Fri, 24 Jul 2015 12:03:56 +0000 (14:03 +0200)]
RTEMS: disable use of CLOCK_MONOTONIC for RTEMS.

Newlib provides CLOCK_MONOTONIC define but actual
(4.11) RTEMS does not provide required support.

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
8 years agoMerge branch 'master' of https://github.com/Vajnar/orte
Michal Sojka [Wed, 10 Jun 2015 22:20:42 +0000 (00:20 +0200)]
Merge branch 'master' of https://github.com/Vajnar/orte

8 years agoJORTE: coding style fixes
Martin Vajnar [Wed, 10 Jun 2015 21:14:35 +0000 (23:14 +0200)]
JORTE: coding style fixes

8 years agotravis-ci: We need newer uncrustify
Michal Sojka [Wed, 10 Jun 2015 16:47:12 +0000 (18:47 +0200)]
travis-ci: We need newer uncrustify

8 years agotravis-ci: Install uncrustify
Michal Sojka [Wed, 10 Jun 2015 16:09:08 +0000 (18:09 +0200)]
travis-ci: Install uncrustify

8 years agoCreate .travis.yml
wentasah [Wed, 10 Jun 2015 15:58:32 +0000 (17:58 +0200)]
Create .travis.yml

8 years agoMerge branch 'shape_fixes' of https://github.com/Vajnar/orte
Michal Sojka [Wed, 10 Jun 2015 12:56:33 +0000 (14:56 +0200)]
Merge branch 'shape_fixes' of https://github.com/Vajnar/orte

8 years agoAndroid: update API levels for building
Martin Vajnar [Wed, 3 Jun 2015 13:56:54 +0000 (15:56 +0200)]
Android: update API levels for building

Build all components against API level 22 (Android 5.1). Decrement
minSdkVersion, because ORTE needs pthread_rwlock_* and these are provided
since API level 9 (Android 2.3).

This is the recommended setting by Google. By saying to target the latest
API, we prevent invoking compatibility behaviors. These behaviors are
sometimes buggy (during testing with 'shape_android' after interacting with
menu items they were duplicated each time I tapped on them).

By saying that minimum API level is 9 we declare, that jORTE does not
require any features of the later APIs and that it runs without any
problems at that level.

I tested 'shape_android' on Android 2.3/4.0/4.1/4.3/4.4/5/5.1.

8 years agoMerge branch 'shape_fixes' of https://github.com/Vajnar/orte
Michal Sojka [Wed, 27 May 2015 08:41:46 +0000 (10:41 +0200)]
Merge branch 'shape_fixes' of https://github.com/Vajnar/orte

8 years agoJORTE: initialize string_pointer variable to make Java compiler happy
Martin Vajnar [Sun, 24 May 2015 21:38:54 +0000 (23:38 +0200)]
JORTE: initialize string_pointer variable to make Java compiler happy

8 years agoshape_android: Stop publishing fresh topics on App destroy
Martin Vajnar [Sun, 24 May 2015 21:30:48 +0000 (23:30 +0200)]
shape_android: Stop publishing fresh topics on App destroy

When application exits stop publishing new data by removing redraw
callback from handler.

8 years agoSHAPE: Fix typo
Martin Vajnar [Sun, 24 May 2015 09:51:13 +0000 (11:51 +0200)]
SHAPE: Fix typo

8 years agoJORTE: document hack that 'bypasses' erratic bug in liborte
Martin Vajnar [Sat, 2 May 2015 13:24:55 +0000 (15:24 +0200)]
JORTE: document hack that 'bypasses' erratic bug in liborte

The recvCallBack() could sometimes be invoked even after subscription's
destruction. In order to avoid dereferencing pointer to free()'d memory we
use this 'intermediary' pointer and during subscription destruction we
set it to null and test this value in recvCallBack().

8 years agoJORTE: add static asserts on values of platform-specific sizes
Martin Vajnar [Sat, 2 May 2015 12:51:10 +0000 (14:51 +0200)]
JORTE: add static asserts on values of platform-specific sizes

If the sizes differ, die with error.

8 years agoJORTE: move common stuff to getString() function
Martin Vajnar [Sat, 2 May 2015 12:36:28 +0000 (14:36 +0200)]
JORTE: move common stuff to getString() function

Create getString() function, that first determines size of char pointer,
then fetches the appropriate number of bytes from the helper buffer and
stores them in an appropriate Java primitive data type. When the pointer
is retrieved native get_string() function is passed the pointer and
it returns Java String object encapsulating the requested C string.

8 years agoJORTE: make helper buffer static
Martin Vajnar [Sat, 2 May 2015 12:22:21 +0000 (14:22 +0200)]
JORTE: make helper buffer static

helper_buffer carries always the same information on a given platform, so
it can be moved to static constructor, where it is initialized only once
during class loading.

8 years agoshape_android: fix freezes on app restart
Martin Vajnar [Sat, 2 May 2015 11:46:43 +0000 (13:46 +0200)]
shape_android: fix freezes on app restart

When application was exited through 'back' button domain associated with
the running instance was destroyed, however the publishers and subscribers
registered within that domain were left intact.

This caused 'Application Not Responding' to be thrown when user wants to
start the application again, because it tries to use handles associated
with the now defunct publishers/subscribers.

In order to prevent this, we destroy Publisher and Subscriber objects upon
invocation of onDestroy() callback method.

8 years agoshape_android: do not enforce ByteOrder
Martin Vajnar [Sat, 2 May 2015 11:38:35 +0000 (13:38 +0200)]
shape_android: do not enforce ByteOrder

Setting of ByteOrder by user is not supported for publications. That is
on the Java part we discover, what the native ByteOrder is and set it
accordingly (see MessageData() constructor at
orte/java/src/org/ocera/orte/types/MessageData.java:41). On the C side it
is assumed, that the byte order is also native
(see ORTEPublicationSendLocked() at orte/liborte/ORTEPublication.c:316).

But if we set the ByteOrder of publisher different than native on the Java
part, than in liborte it is still believed to be native causing problems
during deserialization.

9 years agoJava: rewrite receive callback
Martin Vajnar [Mon, 20 Apr 2015 17:55:29 +0000 (19:55 +0200)]
Java: rewrite receive callback

This commit results in significant increase of subscriber's performance
in Java.

Previously we used to create many Java objects during the reception of fresh
issue. These objects carried the same information as the ORTERecvInfo C
structure. Now we handle it in a way similar to how data reception is taken
care of.

That is, we allocate a direct ByteBuffer upon the creation of subscriber.
This buffer's address is then stored in a helper structure, which is passed
to this receive callback function.

When new issue arrives, we copy it's ORTERecvInfo C structure into this
ByteBuffer by calling memcpy(). We are not immediately parsing the contents
of this buffer for performance reasons. We do it only when someone
requests information included in the buffer (this is handled in the
org.ocera.orte.types.RecvInfo class).

Many method IDs and field IDs previously looked-up upon every callback
invocation are now cached in a helper structure, since the IDs remain
valid until the mentioned classes are unloaded by the ClassLoader
(that is until all their instances are GC'd). Also many unnecessary checks
are removed from the receive callback. This patch increases the probability,
that the callback function remains in the I-cache, because many procedures
are moved away.

Performance comparison before and after applying this patch:
(Reception of 10,000 issues)

PC, Intel C2Q Q9550 @ 2GHz, OpenJDK 7
Before: 2.01 s
Now:    1.68 s

SE Xperia Ray, Android 4.1
Before (Thumb): 10.95 s
Before (ARM):   10.90 s
Now (Thumb):     9.43 s
Now (ARM):       9.53 s

As could be seen the Thumb-2 mode outperforms the ARM mode, so there is now
no need to switch to the ARM mode.

9 years agoAdd orte_config.h for RTEMS
Michal Sojka [Sat, 14 Mar 2015 15:51:02 +0000 (16:51 +0100)]
Add orte_config.h for RTEMS

Since the previous commit, it is no longer possible to compile for Linux
and RTEMS with the same config file.

9 years agoUse CLOCK_MONOTONIC rather than CLOCK_REALTIME
Michal Sojka [Sat, 14 Mar 2015 15:39:35 +0000 (16:39 +0100)]
Use CLOCK_MONOTONIC rather than CLOCK_REALTIME

This changes the clock used internally by ORTE for timestamps in the
protocol and for timing of internal operations. Previously,
CLOCK_REALTIME was used, probably because it is the default clock used
by pthread_cond_timedwait(). Using CLOCK_REALTIME has the disadvantage
that the time can "jump", e.g. when the clock is adjusted by ntpdate.
This causes the applications to stop publishing, which results in them
disappearing from the domain even through their processes still run.

The solution implemented in this commit is to switch the clock to
CLOCK_MONOTONIC, which does not jump. This should have no effect at the
protocol level, because the semantic of timestamps is not defined in the
spec. This was also tested by running the implementations with and
without this change against each other.

In future, it might be a good idea to use different clocks for
timestamps in the protocol and for timing of internal operations (e.g.
ORTEAppSendThread()). I'm not sure what happens in the following two
cases:

1) When one node is suspended to RAM/disk, CLOCK_MONOTONIC stops for the
   time of suspension and it is not clear to me what other nodes that
   run continuously do when the suspended node resumes.

2) When a node is rebooted, the monotonic clock starts from zero, i.e.
   there is a "jump". I'm not sure, whether IDs of the application
   before and after reboot will be different. UDP ports, that are a part
   of the ID should be randomized, but the question is how good this
   randomization is. If the ID is the same before and after reboot some
   problems might also appear.

9 years agoFix linker error on recent MinGW
Michal Sojka [Sat, 14 Mar 2015 14:46:29 +0000 (15:46 +0100)]
Fix linker error on recent MinGW

Commit 5874e67aa7c5fb7b7f82163c1777480c0aaa1541 added -lrt for
compatibility with certain Linux distributions, but on recent MinGW,
this fails.

9 years agoUpdate uncrustify script to be usable in git-filter-branch
Michal Sojka [Sat, 14 Mar 2015 13:15:37 +0000 (14:15 +0100)]
Update uncrustify script to be usable in git-filter-branch

Now, it is possible to use 'git filter-branch --tree-filter orte/uncrustify'.

9 years agoBuild with last Android SDK works with aapt workaround.
jiri hubacek [Mon, 16 Feb 2015 17:10:39 +0000 (18:10 +0100)]
Build with last Android SDK works with aapt workaround.

9 years agoAdd relationship diagram of ORTE data structures
Michal Sojka [Mon, 9 Feb 2015 18:19:57 +0000 (19:19 +0100)]
Add relationship diagram of ORTE data structures

Thanks to Martin Vajnar.

9 years agoUpdate README for shape_android
Michal Sojka [Mon, 9 Feb 2015 17:44:51 +0000 (18:44 +0100)]
Update README for shape_android

9 years agoREADME added to shape_android.
jiri hubacek [Sun, 8 Feb 2015 20:01:25 +0000 (21:01 +0100)]
README added to shape_android.

9 years agoCosmetic modification.
jiri hubacek [Wed, 4 Feb 2015 16:32:17 +0000 (17:32 +0100)]
Cosmetic modification.

- default case in switch()
- space (instead of tab) after TODO

9 years agoWake lock added (prevent device from sleep).
jiri hubacek [Wed, 4 Feb 2015 09:52:43 +0000 (10:52 +0100)]
Wake lock added (prevent device from sleep).

9 years agoFix scaling.
jiri hubacek [Sun, 1 Feb 2015 20:43:24 +0000 (21:43 +0100)]
Fix scaling.

To be dependent only at destination width and height.

9 years agoUse proper types, change messages to Little Endian.
jiri hubacek [Sun, 1 Feb 2015 12:23:18 +0000 (13:23 +0100)]
Use proper types, change messages to Little Endian.

9 years agoFiles which should be omitted.
jiri hubacek [Sat, 31 Jan 2015 21:56:18 +0000 (22:56 +0100)]
Files which should be omitted.

9 years agoRefresh project properties
jiri hubacek [Sat, 31 Jan 2015 21:38:29 +0000 (22:38 +0100)]
Refresh project properties

9 years agoSimplified approach to write application.
jiri hubacek [Sat, 31 Jan 2015 19:51:17 +0000 (20:51 +0100)]
Simplified approach to write application.

- no need for fragments
- no need for complex approach to backward compatibility
- usable drawables choosen
- cs language added
- help added
- settings added

9 years agoMake 'shape4a' compatible with 'shape'.
jiri hubacek [Sat, 27 Sep 2014 19:26:33 +0000 (22:26 +0300)]
Make 'shape4a' compatible with 'shape'.

9 years agoInitial commit of shape4a v1.0 - shape for android.
jiri hubacek [Sat, 20 Sep 2014 07:44:00 +0000 (10:44 +0300)]
Initial commit of shape4a v1.0 - shape for android.

9 years agoReformat the sources with orte/uncrustify script
Michal Sojka [Mon, 9 Feb 2015 16:45:53 +0000 (17:45 +0100)]
Reformat the sources with orte/uncrustify script

9 years agoChange uncrustify configuration according to discussion with Pavel
Michal Sojka [Mon, 9 Feb 2015 16:42:46 +0000 (17:42 +0100)]
Change uncrustify configuration according to discussion with Pavel

9 years agoDon't have a comment between return type and function name
Michal Sojka [Mon, 9 Feb 2015 16:29:57 +0000 (17:29 +0100)]
Don't have a comment between return type and function name

9 years agoAdd configuration for uncrustify
Michal Sojka [Sat, 7 Feb 2015 12:21:22 +0000 (13:21 +0100)]
Add configuration for uncrustify

The formatting of ORTE source code is messy. To unify the formatting at
least a bit, we will use the uncrustify tool. This commit adds
uncrustify configuration tuned for ORTE sources and a helper script that
call uncrustify on all relevant files.

9 years agoFix missing references to clock_gettime for OMK build under Linux.
Pavel Pisa [Tue, 7 Oct 2014 22:43:47 +0000 (00:43 +0200)]
Fix missing references to clock_gettime for OMK build under Linux.

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
9 years agoFix deinitialization in an error path
Michal Sojka [Mon, 6 Oct 2014 23:19:48 +0000 (01:19 +0200)]
Fix deinitialization in an error path

9 years agoFix check-java Makefile target to work with older ant version
Michal Sojka [Mon, 19 Jan 2015 15:50:37 +0000 (16:50 +0100)]
Fix check-java Makefile target to work with older ant version

9 years agoSimplify check target in Makefile.git
Michal Sojka [Mon, 19 Jan 2015 15:30:07 +0000 (16:30 +0100)]
Simplify check target in Makefile.git

9 years agoVarious autoconf updates and fixes
Michal Sojka [Mon, 19 Jan 2015 14:16:01 +0000 (15:16 +0100)]
Various autoconf updates and fixes

9 years agoRemove useless condition
Michal Sojka [Fri, 16 Jan 2015 12:10:13 +0000 (13:10 +0100)]
Remove useless condition

I did small investigation whether the second condition should compare
something else or not and it seems that no other comparison is necessary.
Therefore I believe that removing this rather than changing it to something
else is the correct thing to do.

Thanks to Martin Strunz for spotting this.

9 years agoRTEMS: QEMU provides ETHERNET HW address so there is no need to setup it in application.
Pavel Pisa [Sun, 2 Nov 2014 16:30:20 +0000 (17:30 +0100)]
RTEMS: QEMU provides ETHERNET HW address so there is no need to setup it in application.

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
9 years agoRTEMS: alternative example network configuration for ne2k card under QEMU.
Pavel Pisa [Sun, 2 Nov 2014 16:21:32 +0000 (17:21 +0100)]
RTEMS: alternative example network configuration for ne2k card under QEMU.

This allows to run example build for pc686 board in QEMU
with networking

qemu-system-i386 -kernel _compiled/pc686/bin/orte_rtems_shell_example \
  -net nic,vlan=0,model=ne2k_isa \
  -net tap,script=/etc/qemu-ifup

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
9 years agoRTEMS: use DHCP network configuartion for shell example.
Pavel Pisa [Sun, 2 Nov 2014 16:11:31 +0000 (17:11 +0100)]
RTEMS: use DHCP network configuartion for shell example.

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
9 years agoFunction ReleaseStringUTFChars() needs jstring as 2nd argument instead of PublProp...
jiri hubacek [Sun, 21 Sep 2014 10:12:34 +0000 (13:12 +0300)]
Function ReleaseStringUTFChars() needs jstring as 2nd argument instead of PublProp/SubsProp object.

9 years agoROBOT_DEMO: reduce WifiLocks usage
Martin Vajnar [Tue, 8 Apr 2014 11:48:31 +0000 (13:48 +0200)]
ROBOT_DEMO: reduce WifiLocks usage

Use WifiLock only for MotionSpeedPublisher.

9 years agoJORTE: fix bug in subscriber callback
Martin Vajnar [Tue, 8 Apr 2014 08:51:27 +0000 (10:51 +0200)]
JORTE: fix bug in subscriber callback

This fixes a rarely occurring bug in subscriber callback.

When a Java subscriber is being destroyed, during the process
a JORTECallbackContext_t structure, used for storing pointers to Java objects,
was free()'d. If this occurred when new issue had arrived and was being
processed by the receive thread, then the pointer to the, now non-existent,
callback structure would have been dereferenced, leading to a SEGFAULT.

To fix this we now create a pointer to pointer to the specific structure
and pass this pointer to ORTESubcriptionCreate() and use it from the recvCallBack().

9 years agoRoboDruid: use Toasts for short messages
Martin Vajnar [Tue, 4 Feb 2014 19:57:42 +0000 (20:57 +0100)]
RoboDruid: use Toasts for short messages

9 years agoRoboDruid: make release build
Martin Vajnar [Sun, 2 Feb 2014 12:37:44 +0000 (13:37 +0100)]
RoboDruid: make release build

9 years agoRoboDruid: organize classes into packages
Martin Vajnar [Sat, 1 Feb 2014 17:19:11 +0000 (18:19 +0100)]
RoboDruid: organize classes into packages

9 years agoRoboDruid: improve handling of short taps
Martin Vajnar [Thu, 23 Jan 2014 22:19:36 +0000 (23:19 +0100)]
RoboDruid: improve handling of short taps

9 years agoRoboDruid: toggle Hokuyo output and speed control/monitor by touch input
Martin Vajnar [Wed, 22 Jan 2014 23:19:14 +0000 (00:19 +0100)]
RoboDruid: toggle Hokuyo output and speed control/monitor by touch input

Single tap on Hokuyo output starts/stops subscriber.
Single tap on speed vector starts/stops subscriber,
long tap starts/stops publisher.

9 years agoDocument gAVL, linked list and queue declaration
Martin Vajnar [Sun, 11 May 2014 18:51:50 +0000 (20:51 +0200)]
Document gAVL, linked list and queue declaration

9 years agoAdd missing FREE()
Michal Sojka [Mon, 5 May 2014 12:18:00 +0000 (14:18 +0200)]
Add missing FREE()

Thanks to Martin Strunz <martfi@aol.com> for noticing
this.

10 years agoDocument gAVL declaration
Michal Sojka [Tue, 15 Apr 2014 08:38:38 +0000 (10:38 +0200)]
Document gAVL declaration

10 years agoRTEMS: actual RTEMS git sources do not provide CONFIGURE_NUMBER_OF_DRIVERS define.
Pavel Pisa [Sun, 13 Apr 2014 11:17:08 +0000 (13:17 +0200)]
RTEMS: actual RTEMS git sources do not provide CONFIGURE_NUMBER_OF_DRIVERS define.

There is only one version of table - actual runtime version.
The read-only table with initialization data is not present
and cannot be used for to compute number of entries of config
harcoded drivers set (CONFIGURE_NUMBER_OF_DRIVERS).

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
10 years agoRTEMS: allocate space for POSIX keys which are required by current RTEMS shell version.
Pavel Pisa [Sun, 13 Apr 2014 11:07:03 +0000 (13:07 +0200)]
RTEMS: allocate space for POSIX keys which are required by current RTEMS shell version.

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
10 years agoInitial version of Debian packaging
Michal Sojka [Sat, 15 Mar 2014 22:49:23 +0000 (23:49 +0100)]
Initial version of Debian packaging

10 years agoorte-idl: Fix possible security problem
Michal Sojka [Sun, 16 Mar 2014 00:40:03 +0000 (01:40 +0100)]
orte-idl: Fix possible security problem

10 years agoUpdate versioning of shared libraries
Michal Sojka [Sat, 15 Mar 2014 23:48:48 +0000 (00:48 +0100)]
Update versioning of shared libraries

Previously, we used -release to specify the SONAME, which would cause
binary incompatibility between releases even if the library is not
changed at all.  Additionally, libtool interface numbers were derived
from relase number in an undocumented way. This is removed as well and
now we have to modify SONAME manually, when the interface changes.

10 years agoRevert "Compile shared library by default"
Michal Sojka [Tue, 4 Mar 2014 15:45:50 +0000 (16:45 +0100)]
Revert "Compile shared library by default"

This reverts commit 9c71dc94e674c0ef52efabcd3dbddebceb2dc5f4.

OMK is broken and some things do not work when shared libraries are
created. This has to wait for later. If you need shared library, use
automake-based build.

10 years agoCompile shared library by default
Michal Sojka [Tue, 4 Mar 2014 15:37:15 +0000 (16:37 +0100)]
Compile shared library by default

10 years agoFix reported problem "compile error on cygwin"
Pavel Pisa [Thu, 14 Nov 2013 13:02:50 +0000 (14:02 +0100)]
Fix reported problem "compile error on cygwin"

The Windows service support in ORTE manager is enabled
now for Cygwin build as well. Build under MINGW has been
tested but some external volunteer is required for Cygwin
environment test.

Another option is to add config option which allows
to specify if Windows service support should or should not
be build.

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
10 years agoExtend and correct RTEMS instructions README file.
Pavel Pisa [Thu, 14 Nov 2013 12:01:58 +0000 (13:01 +0100)]
Extend and correct RTEMS instructions README file.

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
10 years agoDocument and made more straightforward RTEMS build.
Pavel Pisa [Thu, 14 Nov 2013 11:59:18 +0000 (12:59 +0100)]
Document and made more straightforward RTEMS build.

The local copy of OMK Makefile.rules for RTEMS
is included

  Makefile.rules.rtems

and RTEMS build is documented

  README.rtems

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
10 years agoBump version to 0.3.4 v0.3.4
Michal Sojka [Mon, 28 Oct 2013 21:33:52 +0000 (22:33 +0100)]
Bump version to 0.3.4

10 years agoAdd 0.3.4 release notes
Michal Sojka [Mon, 28 Oct 2013 21:29:44 +0000 (22:29 +0100)]
Add 0.3.4 release notes

10 years agoFix dist makefile target
Michal Sojka [Mon, 28 Oct 2013 21:29:04 +0000 (22:29 +0100)]
Fix dist makefile target

10 years agoUpdate README.windows
Michal Sojka [Mon, 28 Oct 2013 21:05:08 +0000 (22:05 +0100)]
Update README.windows

10 years agoAdd checks for omk builds
Michal Sojka [Mon, 28 Oct 2013 20:58:58 +0000 (21:58 +0100)]
Add checks for omk builds

10 years agoFix check target
Michal Sojka [Mon, 28 Oct 2013 20:41:46 +0000 (21:41 +0100)]
Fix check target

10 years agoFix the check for Java build
Michal Sojka [Mon, 28 Oct 2013 20:39:55 +0000 (21:39 +0100)]
Fix the check for Java build

10 years agoDo not use ugly directory names for Java build results
Michal Sojka [Mon, 28 Oct 2013 20:38:20 +0000 (21:38 +0100)]
Do not use ugly directory names for Java build results

10 years agoUpdate of local copy of Windows pthreads library to MinGW-w64 winpthreads.
Pavel Pisa [Mon, 28 Oct 2013 20:46:36 +0000 (21:46 +0100)]
Update of local copy of Windows pthreads library to MinGW-w64 winpthreads.

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
10 years agoAdd newly tested platforms
Michal Sojka [Mon, 28 Oct 2013 20:13:02 +0000 (21:13 +0100)]
Add newly tested platforms

10 years agoAdd android build check
Michal Sojka [Mon, 28 Oct 2013 20:04:51 +0000 (21:04 +0100)]
Add android build check

10 years agoRemove 'cd' from check commands
Michal Sojka [Mon, 28 Oct 2013 19:52:45 +0000 (20:52 +0100)]
Remove 'cd' from check commands

10 years agoRemove dist from OMK makefile
Michal Sojka [Mon, 28 Oct 2013 19:42:24 +0000 (20:42 +0100)]
Remove dist from OMK makefile

OMK leaks unwanted environment variables to the Makefile.cvs.

10 years agoPerform compilation checks with --enable-orte-idl
Michal Sojka [Mon, 28 Oct 2013 19:28:13 +0000 (20:28 +0100)]
Perform compilation checks with --enable-orte-idl

10 years agoUpdate README.android
Michal Sojka [Mon, 28 Oct 2013 19:27:29 +0000 (20:27 +0100)]
Update README.android

10 years agoMove robot demo to contrib/
Michal Sojka [Mon, 28 Oct 2013 19:19:36 +0000 (20:19 +0100)]
Move robot demo to contrib/

10 years agoCheck whether windows dll is produced
Michal Sojka [Mon, 28 Oct 2013 16:22:36 +0000 (17:22 +0100)]
Check whether windows dll is produced

10 years agoROBOT_DEMO: use new constants and fix behavior of callback functions
Martin Vajnar [Sun, 20 Oct 2013 16:07:55 +0000 (18:07 +0200)]
ROBOT_DEMO: use new constants and fix behavior of callback functions

Test the receive status to see whether the callback function was called
because of new new data reception or because of a deadline expiration.