Commit Graph

5599 Commits

Author SHA1 Message Date
James Turner
6167159795 API to reset the shared tree geometry 2020-08-26 10:37:54 +01:00
James Turner
672afdbc34 Fix for sprintf of nil numeric in Nasal
Spotted by valgrind, w00t. If you used a numerical formatter, such
as f,d or g, with. a nil value, we would read uninitialised data, and
generally get confused.
2020-08-26 10:37:54 +01:00
James Turner
27e61b3dec Ensure SGTerraSync::unbind drops all properties
Avoid some noise on reset, by dropping /all/ our SGPropertyNodes
2020-08-26 10:37:54 +01:00
legoboyvdlp R
f1d00c9b40 Drizzle is considered to be light rain. Backportable to 2020.2. Fixes ticket 1792 2020-08-26 10:37:54 +01:00
James Turner
1b00ece8c4 SGBuildingBin: avoid read of un-inited memory
Where the building line is not as long as expected, ensure we read
valid memory.
2020-08-26 10:37:54 +01:00
James Turner
d0f24229b2 HTTP: Use curl_multi_wait everywhere
Also check the result of the curl methods, in case they fail
2020-08-26 10:37:54 +01:00
Richard Harrison
ab8795f6dc Directional lighting fix
getLights modified to use points rather than triangles based on the configuration.

This fixes point lights (with directional disabled) on AMD
2020-08-26 10:37:42 +01:00
James Turner
2fe60c9635 Terraysnc: fix incorrect handling of ‘entirely ocean’ tiles
We had some confusion between two error codes, meaning we would log an
error (and treat as a failure) a pure ocean 1x1 area, even though this
was entirely correct and expected behaviour.

Terrasync.cxx expected REPO_ERROR_NOT_FOUND, we need to convert the
file-level error to that, and avoid logging a warning.
2020-08-25 21:10:54 +01:00
James Turner
27a3ee3bce TerraSync: restart after max-errors is exceeded.
When we trip the max-error count for a session, back off for a period
of time and then retry (selecting a new TerraSync server).
2020-08-25 21:02:39 +01:00
James Turner
0721db3acd TerraSync: handle reinit better
Fix various cases where re-init could get things blocked. Remove the
duplicate storage of the active paths; now we always check the primary
data, and hence it can’t be out of sync.

Also remove the obsolete persistent cache code.

Fixes some of the issues discussed in:
https://sourceforge.net/p/flightgear/codetickets/2308/

Further improvements still to come, especially to retry on a better
schedule for intermittent connections.
2020-08-25 21:02:16 +01:00
Richard Harrison
24b58cbe21 XMLSound bugfixes
- using expressions for pitch and volume didn't work
- multiple volume elements not computed properly

Discussion here: https://web.archive.org/web/20200730232940/http://fguk.me/forum/development-hangar/8360-supersonic-audio?start=40#46287

          <volume>
                <property>/a/v1</property>
                <factor>0.1</factor>
            </volume>
            <volume>
                <property>/a/v2</property>
                <offset>1</offset>
                <factor>-3</factor>
            </volume>
            <volume>
                <property>/a/v3</property>
                <offset>1</offset>
                <factor>-1</factor>
            </volume>

evaluates as follows:

update(): Have 3 volume entries
 --> 0:prop /a[0]/v1[0] => 0.110000 factor = 0.100000 v=0.011000 offset 0.000000, now 0.000000, v=0.011000 vol=0.011000 ==> 0.011000
 --> 1:prop /a[0]/v2[0] => 0.120000 factor = 3.000000 v=0.360000 -ve offset 1.000000 1.360000 = 0.014960  ==> 0.014960
 --> 2:prop /a[0]/v3[0] => 0.130000 factor = 1.000000 v=0.130000 -ve offset 1.000000 1.130000 = 0.016905  ==> 0.016905
2020-08-04 11:58:27 +01:00
Richard Harrison
5b3274e688 Exclude images with less than 16 bits per pixel from the DDS Texture Cache 2020-08-04 11:58:13 +01:00
James Turner
fcd75cfae5 Logging: add log overload which copies filename
This is needed to fix a Valgrind error in the test-cases, due to logging
Nasal filenames which are GCed.
2020-08-04 11:58:05 +01:00
James Turner
6387a1d6d0 Fix compilation for some MSVC versions 2020-07-14 13:59:14 +01:00
Julian Smith
fc4ce2528b Renamed version -> simgear-version to avoid breaking clang++ on OpenBSD.
It seems that clang++ headers #include <version>, which found simgear/version
because we need to put singear/ in include path for some code to compile.
2020-07-14 13:59:09 +01:00
James Turner
18d2bfcd8b HTTP: allow CAINFO to be set
Env var is SIMGEAR_TLS_CERT_PATH
2020-07-14 13:58:43 +01:00
James Turner
6738a3aa2b Helper to map string to log priority 2020-06-26 10:31:16 +01:00
Julian Smith
a8c1bef0bf Avoid various gcc warnings. 2020-06-26 10:30:58 +01:00
James Turner
4faf0ea468 API to reset the resource manager, and remove
providers.

Needed to fix a bug with parsing -set.xml files in the launcher;
correctly loading XML needs the resource paths to be defined.
2020-06-26 10:30:47 +01:00
Lars Toenning
80cc09fe90 Fix compiler warning of different signedness 2020-06-26 10:30:37 +01:00
James Turner
ca8dbb985e Fix zero-interval repeat on pick/knob animation
See:
https://sourceforge.net/p/flightgear/codetickets/2241/

Note did not adjust bug in knob-animation ignoring repeatable flag
(always marked repeatable) since this might break some aircraft.
2020-06-15 16:13:39 +01:00
James Turner
751cdc32a8 new version: 2020.2.1 2020-06-15 12:33:50 +01:00
Julian Smith
1d978429f5 simgear/props/props.cxx: Avoid possible SEGV e.g. if listener removes itself.
simgear/props/props_test.cxx: test for self-unregister.

SGPropertyNode::removeChangeListener() used to modify the _listeners vector or
even delete it and set _listeners to NULL, if a listener was removed.

This could cause a SEGV if removeChangeListener() was called from
a listener, because listeners are called from methods such as
SGPropertyNode::fireValueChanged() which iterate through _listeners and don't
expect it to be changed to NULL.

The fix here is to keep a track of whether we are iterating through the
_listeners vector; if we are, SGPropertyNode::removeChangeListener() sets the
entry to NULL, instead of removing it.

We have a new internal function forEachListener() which takes
care of the iterating details - incrementing/decrementing the new
_listeners->_num_iterators count and skipping NULL entries. Before returning,
if _num_iterators is now zero it removes any NULL entries in the vector of
listeners.

The nListeners() member is no longer inline, because it needs to ignore NULL
entries in the vector. And _listeners now points to an internally-defined
struct which contains a new _num_iterators as well as the vector of listeners.

Added self-unregister test to simgear/props/props_test.cxx.
2020-06-13 18:11:49 +01:00
James Turner
2b2e3ae5c4 Fix CMake value 2020-06-05 13:53:54 +01:00
James Turner
0e3ac7e078 Set CMake OpenGL VND policy 2020-06-05 13:32:43 +01:00
James Turner
01ab3b2385 New SGGeodesy::advance, taking degrees 2020-05-30 16:27:42 +01:00
James Turner
11cf87951c SIMD flags: ensure we respect other compiler flags
Ensure we only ever extend the currently set compile flags when
ENABLE_SIMD is requested
2020-05-28 10:46:18 +01:00
James Turner
7c004a4c90 Fix compilation with Boost >= 1.73 2020-05-28 10:45:47 +01:00
James Turner
a48693d273 Kill explicit setting of compiler flags in Debug
Use the CMake defaults for this.
2020-05-24 14:32:13 +01:00
James Turner
12d57a6373 Adjust RelWithDebInfo optimisation for Clang
This should give a bit of speed boost to macOS official builds.
2020-05-13 14:23:00 +01:00
James Turner
c7c8fc7777 Better RelWithDebInfo flags for CL.exe 2020-05-08 14:31:44 +01:00
James Turner
6e054e57ef Windows: ensure RelWithDebInfo is fast
Default CMake RelWithDebInfo pessimizes inlining
2020-05-07 14:13:54 +01:00
James Turner
1d89a76d13 Old-style texture compression: better check
Make even more sure new-style texture-compression is active, before
deciding which old-style mode to use.
2020-05-07 11:47:25 +01:00
James Turner
65925cccdf Enable old-style texture compression
Thanks to Dany for tracking this down!
2020-05-05 20:40:29 +01:00
James Turner
898559ab31 Add copyPropertiesIf helper 2020-05-03 12:21:36 +01:00
James Turner
ad3621e23b Add Canvas::Image.dirtyPixels() 2020-04-28 09:36:40 +01:00
James Turner
57b0f70a55 Nasal: improve message for non-object member access
Based on some discussion on this ticket:

https://sourceforge.net/p/flightgear/codetickets/2186/

Make this message slightly clearer.
2020-04-28 09:36:40 +01:00
Erik Hofman
8d6e543cc2 Only use readwav for OpenAL 2020-04-28 09:41:11 +02:00
James Turner
bd93fb279b new version: 2020.2.0 2020-04-27 09:29:55 +01:00
James Turner
81bdae283a new version: 2020.1.1 2020-04-27 09:29:54 +01:00
James Turner
00e25404c7 Add simgear::optional
This is a poor-man’s proxy for std::optional until we have it.
2020-04-25 18:05:59 +01:00
gallaert
6a157e1c08 Replace boost::tuple by std::tuple 2020-04-22 14:04:21 +01:00
gallaert
7a6f9151a6 General boost cleanup 2020-04-22 14:04:21 +01:00
gallaert
567293c14d Replace boost::function by std::function 2020-04-22 14:04:21 +01:00
gallaert
2ff6a4966a Replace boost::bind by std::bind and lambda function 2020-04-22 14:04:21 +01:00
gallaert
b13d7f9392 Replace boost::lexical_cast by std::ostringstream and std:stof/stoi/stod 2020-04-22 14:04:21 +01:00
gallaert
d21137739d Replace boost::enable_if/disable_if/enable_if_c/disable_if_c by std::enable_if 2020-04-22 14:03:47 +01:00
gallaert
1e13cf6fec Replace boost::shared_ptr/weak_ptr by std::shared_ptr/weak_ptr 2020-04-22 14:03:47 +01:00
James Turner
5c52cdc255 Fix for crash on reset:
https://sourceforge.net/p/flightgear/codetickets/2217/

Ensure that subsystem lookup during its own removal, returns nullptr.
Otherwise we cache an about-to-be dead pointer, which is Bad (TM)
2020-04-22 10:55:41 +01:00
James Turner
17d7808200 Remove a log message 2020-04-22 10:54:44 +01:00