mirror of
https://github.com/MaSzyna-EU07/maszyna.git
synced 2026-11-21 06:14:58 +01:00
201cae07e15509dd255caf892a75a603a45649f7
3704 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
|
201cae07e1 |
fix: a failed start says why and ends with an error code; a small crash dump by default (#425)
* fix: a failed start says why in the log and ends with an error code When the window could not be made, the log only said "Bad init: failed to create glfw window", without the reason glfw knows (no OpenGL driver, no pixel format that fits, no EGL). Sometimes it did not even say that: the log thread writes every 50 ms, and main() ended the process with std::_Exit(0) without stopping it, so a start that failed within that time left a log that ends at "// startup". - glfw's errors go to the log and errors.txt through glfwSetErrorCallback, set before glfwInit(); each distinct message once. - after a failed start (or an exception caught in main) the log thread is stopped and waited for, as the test runs already do, so nothing logged last is lost. - a failed start ends with exit code 1 instead of 0, so the starter can tell it from a closing. A start that worked ends as before. * fix: glfw's errors told during the start only, a closed crash dialog is no failed start After review of the previous commit: - the glfw error callback is taken off at the end of init(). Once the simulator runs, glfw would only repeat what the platform can't do (glfwSetCursorPos on Wayland, on every mouse-look) or what a scenery asks for in vain (launcher key codes below GLFW_KEY_SPACE, checked on every update), which would put new lines in errors.txt of existing sceneries and cost a format and a lock per call. - closing the crash-report window ends the simulator as before, but init() now returns 1 for it (above 0: closed by the player), and only a result below 0 gives exit code 1. - a start that throws from init() (bad_alloc, or a runtime_error on Windows) ends with exit code 1 too; an exception after a start that worked ends with 0 as before. * feat: a small crash dump by default, the full one with "crashdump.full yes" The dump written by the crash handler (Windows, WITHDUMPGEN) held all the memory of the simulator (MiniDumpWithFullMemory), so it was as large as the memory the simulator took: gigabytes on a large scenery, written to the game folder at the moment of the crash, on machines with 8 GB of RAM too. By default it now holds the stacks with the memory they point to, the threads, the modules, the handles, the memory map and the modules' data segments (MiniDumpWithDataSegs in place of MiniDumpWithFullMemory, the rest of the flags as before). The globals, Global among them, stay readable. "crashdump.full yes" in eu07.ini writes the full dump as before. Measured on td.scn with a ghostview camera (1.2 GB committed): the small dump 44 MB in 2 s, the full one 1.6 GB in 19 s. The key is new and optional: eu07.ini files without it read as before, and the setting is listed in the log at start. * fix: the small crash dump keeps the process and thread blocks MiniDumpWithProcessThreadData adds the PEB and the TEBs, a few KB: the command line (the scenery and the vehicle), the environment and the last error of each thread, which the small dump otherwise lacks, since the heap behind Global's longer strings is not in it. * fix: an exception that ends the simulator says what it was in the log After review of #425: - the exceptions main() catches are logged ("Critical error: ..."): on Windows the runtime_error that only showed a message box, and on the other systems any std::exception, which used to end the simulator through std::terminate with no word in the log. On Windows the others are still left to CrashHandler, whose dump shows where they were thrown. - every exception caught there ends with exit code 1, in the start or later, as a crash; it used to be 0 when caught, the abort code when not. - closing the crash-report window logs "Crash report window closed, closing" instead of the error "Failed to run crash gui!", which stays for a result below 0. * feat: the stack of a crash on Linux in the log and on the standard error On Linux no dump is written (the crash handler is Windows only, crashpad is off), so a crash left a log that just stopped. A handler of SIGSEGV, SIGBUS, SIGFPE, SIGILL and SIGABRT (installed at the start of main) now writes, with write() only: Crash: signal 11 (SIGSEGV) at address 0x... Version: EU07 (cmake), <hash> (...) Stack (addr2line -f -C -e <the eu07 file of this version> <offset>): <backtrace_symbols_fd> to the end of the log file (its whole path, set when the log opens it) and to the standard error, which StarterNG shows first. The Linux builds carry their symbols (GENERATE_PDB adds -g, nothing strips them), so addr2line on the eu07 file of the same version gives the files and lines. Then the signal ends the process as before (SA_RESETHAND, raise): the same exit status, and a core dump where the system makes one. The handler runs on an alternate stack, so an overflow of the main thread's stack is reported too; backtrace() is warmed up at install so it doesn't allocate in the handler, and alarm(5) ends a handler that hangs (a crash in the allocator or the loader). WAMR is built without its hardware bound checks, so it installs no SIGSEGV handler of its own. Elsewhere crash_signals does nothing. A unit test (Linux only) crashes a child process and checks the report on its standard error and in the log, and that the child still ended by SIGSEGV. * fix: the crash report can't be cut by a second crash signal or end with SIGALRM After review: - the five crash signals are blocked while a report is written, so a second one waits rather than starting a second report in its middle (a fault that comes while blocked still ends the process). - a report that hangs past alarm(5) ends the process by the signal of the crash, not by SIGALRM: a SIGALRM handler, set only in the crash handler, raises the crash signal and unblocks it. - the child of the unit test calls alarm(10) first, so a child stuck before the crash can't hang the build. --------- Co-authored-by: maj00r <maj00r@host> |
||
|
|
ba372e9830 |
Editor: switch drives with no switch listed and marked; a deleted switch takes its drives and ballast along (#330)
* feat(track): the rail code split by concern, the blades of the switches with the key frogs shaped from their templates The rails are laid by code of their own instead of inside Track.cpp: gauge lines (GaugeLine), knot lines (Knots.h), crossings and check rails (RailCrossings), rail materials (RailMaterials), measuring the rail profile (RailProfile), planning the cuts (RailPlan), the carving scan (RailCarve) and laying the geometry (RailLayer). A switch with the key frogs draws its blades from the tip where their planed nose ends to the heel, bent as they're thrown, and the rail behind them from the heel on, where they were laid from the start of the switch before; on a switch of the catalogue they start at the cut tip of the tables, 1.248 m from the joint at R 190, 1.5 m at R 300 (track_features::blades). The frogs cut from the rails, the wing and the check rails are built but switched off (track_features::crossings), the rails laid whole as before. The switches without the key keep the blades they had. A switch path can be a chain of pieces: pieces1 and pieces2 keep the parts of each path after the classic bezier, which stays as the approximation older versions read. A path with its radius given is a curve however gently it bends, a plain track a straight only where both of its control vectors lie along it. A vehicle trailing through a switch with a movable nose set the other way runs off the rails at it (TRAILED_SWITCH). * feat(render): the shared treads of the rails and their cut faces drawn, the cut faces in the matte variants of the rail materials Both renderers draw the treads a track shares with the rails it crosses, in the pick pass too, and a pass of their own for the faces where the rails are cut, the ends of the blades among them, in the matte variants of the rail materials, so the cut faces don't shine like the head the wheels polish. * feat(vehicle): an axle on a switch keeps to the path it came on, and coming from the frog end trails the switch through where the wheels reach the blades The switch thrown under an axle no longer moves it over onto the other path; it runs on along the one it entered. Coming from the frog end, where the blades are shaped or the frog has a movable nose, the switch set the other way is trailed through (rozprucie) only where the wheels get to the blades or to the nose, not as the axle enters the track; the other switches are trailed through as it enters, as before. * fix(track): the tracks drawn straight before, which their second control vector bends now, written to the log The simulator told a straight by the first control vector of a path alone, so a track whose second control vector bent it was drawn straight; now it's a curve. Such a track is named in the log as it loads, so a scenery made counting on the straight can be found: 25 tracks of lk357. Paths with both control vectors along them bend a centimetre or so at most and aren't named. * feat(track): 3D sleepers and switch timbers laid where the simulator makes the ballast, switched off in the build The sleepers of a track running alone are laid square to it, the timbers of the switches and crossings across the tracks running over them, shared by them, of the models set with sleepers3d and sleepers3dplain in eu07.ini; sleepers3dbed tells whether the ballast is laid under them too, lowered so they stick out of it. The layout of a cluster of tracks is worked out once and kept till a track of it changes, in a thread of its own while the simulation runs. timbermodel of a track, read and written back before, takes the place of the model of sleepers3d there. All of it is off whatever eu07.ini sets (track_features::sleepers), the tracks keeping their ballast and their sleeper models. * feat(render): the 3D sleepers and switch timbers drawn in instanced batches Render_Timbers lays the beams of the tracks in range where they're out of date and draws those of a run of tracks of a cell gathered by the model and its part or its level of detail, in as few instanced batches as they fit in; in the shadow pass only those near the camera. A track with them leaves its old sleeper model out. OpenGL 3.3 renderer only. * fix(track): the counts of what's wrong with the sleepers kept safe from the thread laying them, and that thread stopped before what it works with goes The counts of the beams clipped, dropped and off the plane, of the slowest and the biggest cluster, are counted by the thread working out the layouts in the background as well as by the simulation laying a cluster itself, and were plain values written by both; they're atomic now. What the thread keeps the clusters in, their version and the counts were made after the worker, as it first asked for them, and so destroyed before it at the exit while its thread could still be laying a cluster; they're made first now, the worker stopping and joining its thread before they go. The tracks go with the paths of the simulation, made before the worker. * revert(track): a tributary and a track of an unknown type load as before They came in with the rail code from the guards of the editor against crashing, and don't belong with the blades: a track of an unknown type loaded as a normal track of the railway could be joined to it and run on, and a tributary loaded as a river with its first path lost the second one when the editor saved the scenery. Both load as on master again; the place where the editor crashed on them is to be guarded where it is. * fix(track): the tracks drawn straight before named in the log once, as they load, and the paths of the switches with no radius told a straight as before in one term The log of the tracks the second control vector bends now was written each time the segments of a track were made again, as a track with its unevenness loads and as the editor changes or drags a track; it's written only the first time, as the track loads. The paths of a switch or a crossing with no radius keep the rule of master, a straight told by the first control vector alone, its two terms being the same length; they're one term now, with the comment saying so. * fix(track): the thread laying the sleepers stopped as the application exits, and loading no models itself The thread reads the scenery, which the application deletes as it exits well before the statics go: it's stopped and joined first thing in exit() now, the worker destroyed with the statics only where the application exits some other way, and not started again after. The models the beams are made of were loaded by the thread where it didn't find them, through the renderer, out of the thread of the renderer, into a map read by the simulation meanwhile: they're found under a lock now and loaded only by the threads other than it, again for all the tracks whenever a track changed since, before a layout is asked of it. With the 3D sleepers off in the build their settings aren't exported, and a track changed doesn't look for the tracks around to lay their beams again. * feat(track): the 3D sleepers turned on with sleepers3d in eu07.ini, no longer switched off in the build track_features::sleepers is gone: sleepers3d <model> in eu07.ini lays them, none (the default) lays nothing and the tracks keep their ballast, as before. The sleepers3d settings are written back with the others. The editor changes the tracks, so while it's open their sleepers are worked out on the main thread as they're needed, not in the background. * fix(render): only the part of a timber model drawn, not what else the model holds A timber model of parts may hold more than its begin, main and end, as the kits with the fastenings of the rails do. Those were drawn with each part of each timber, in the middle of the track; they aren't laid, so they're skipped now. * fix(track): the 3D sleepers laid only for a renderer which draws them Only the OpenGL 3.3 renderer draws the beams. With another one a track set to them lost its ballast to them, lowered or not laid at all with sleepers3dbed no, and nothing was drawn in its place. Draws_Timbers() of the renderer tells now, and where it's false the tracks keep their ballast, as with sleepers3d none. * fix(render): the submodels of a timber model left as they were after its parts are drawn They were all made visible after a part was drawn, the ones the model hides itself too, like Light_On, while the same model may stand in the scenery. The submodels of the top level besides the parts are listed as the model is loaded, hidden while a part is drawn and given back their own visibility after. The parts have to be at the top level of the model for that; a model with them deeper is logged and drawn whole. Render_Timbers no longer looks the model up under the lock for every run of tracks, the name set is enough there. * fix(render): a vertex whose faces give its texture no direction gets a tangent across its normal, not a NaN one The cut ends of the rails have one spot of the texture on all their corners, so their tangent came out of normalize() as NaN. The normal-mapped shaders then shaded them with a NaN normal: about half of the vertices of the cuts and frogs, and the end faces of the rails. * feat(track): the frogs cut from the rails turned on with frogs in eu07.ini, no longer switched off in the build frogs yes cuts the rails of the plain tracks and of the switches with the key frogs where they cross other rails, into frogs, wing rails and check rails. It's off by default: the rails are laid whole as before, the blades of the switches with the key are shaped still. * refactor(editor): the track mode file split into the files of its tools, the code as it was A pure move, so that the commits after it show what changes in each tool rather than a 4800-line file rewritten. The member functions of editor_mode go to the file the editor of this branch has them in: route, straights, lay, switchplace, switchdrive, switchtool, trackgizmo, and the rest stays in editormodetrack.cpp. The helpers of its anonymous namespace go to editormodetrackshared.h, as they were but inline, in namespace track_mode, which the files use. No line of code changes; the files add only their license, includes and the using line. * feat(editor): the groundwork the tools of the editor share - The tool interface (editortool.h), the gestures of the mouse (editorgestures.h) and what the modes share (editormodeshared.h); the projection, viewport, selection and texture picking taking them up. - Scripted input (scriptedinput): the input of a selftest script fed to the editor as if typed and clicked, the real mouse and keyboard ignored meanwhile; EU07_EDITOR_SELFTEST_HIDDEN draws into a window never shown nor focused. - The ui layers and panels of the editor, its settings, icons and the info of the includes; the layers of the scenery and their saving; the launcher mode. - utilities: user_cache_path (the cache folder of the system for the user) and the helpers of the translations. - stb_image: the gif loader takes the frame two back from the frames it decoded, not from before its buffer. - tools/editor: the scripts making the headers of the drives, the turntables, the editor and its usage. - Unit tests: test_area and the stubs of the editor the tests link. Not built on its own: the tree builds again with the selftests commit, the one before the translation. * feat(editor): the tracks laid along an alignment, to the PKP PLK norms, with their grade line and the ground - The alignment of the laying (editorAlignment, editorAlignmentStore): straights, transition curves and arcs with the cant of the norms for the speed, the radius below the minimum for it a warning; the lay, route and straights tools build on it, the route plan reading the curves going on from the tracks adjoining. - The grade line (editorProfile, the profile and gradeline tools): fitted to the terrain model or edited by hand, the tracks fitted marked in the scenery (//$f), the shared grade lines of nearly parallel tracks (//$n). - The ground under the tracks (editorGroundFollow, trackbed): the terrain drawn to the tracks and restored, the tracks put onto the terrain. - Connections between tracks (editorConnections, the connection and connection curve tools), the track presets, the shape of a track (editorTrackShape), digitizing tracks from the orthophoto, the properties sheet of the tracks (trackprops) and the track ux. - Unit tests: alignment, ground fit, heights, with the data of two fits. Not built on its own: the tree builds again with the selftests commit, the one before the translation. * feat(editor): the switches placed, slid and designed from their templates - The switch templates (editorSwitchStore, the template tool): the catalogue switches and the user's own, kept with the settings; the designer (editorSwitchDesign) with a drawing of the dimensions, a and b typed in with the shape making up the rest, a switch clicked off the orthophoto. - The curve of a switch laid into a curve (editorSwitchCurve), placement on a straight or into a curve, slide along it, crossovers and double slips (switchplace, switchslide), the drives of the switches (switchdrive) and the records of the switches placed (records). - Unit tests: test_switch_curve. Not built on its own: the tree builds again with the selftests commit, the one before the translation. * feat(editor): the tools of the infrastructure beside the tracks The gauge, speed, joints, lineside, parallel tracks, vehicles, signals, turntables, sweep and road tools with headers of their own; the kilometrage of the lines (editorKilometrage, the km tool); the infrastructure, roads and speeds behind them (editorInfra, editorRoad, editorSpeed). Not built on its own: the tree builds again with the selftests commit, the one before the translation. * feat(editor): the terrain model of geoportal.gov.pl and the orthophoto downloaded for the editor - editorNmt: the squares of the terrain model at 1 m (NMT GRID1, the WCS service of geoportal.gov.pl) fetched by two workers, kept in editor/cache/nmt/grid1 of the game folder, the heights of the scenery taken from them; the nmt mode loads it along the tracks or around the camera. - editorHttp: the HTTP client of both, WinHTTP on Windows, libcurl elsewhere where it's built with it; without it the caches only. - The orthophoto: its cache moved to the cache folder of the system for the user, the download area a polygon on the topo map (//$a) filled in the background, the 4K imagery, the tiles waiting outlined on the ground. - editorGeoMap: the topo map of the top view; heightmapterrain and the terrain streamer grow the terrain made from the model around the camera. Not built on its own: the tree builds again with the selftests commit, the one before the translation. * feat(launcher): the scenery wizard picks the place of a new scenery on the topo map the editor shares The wizard drops its own map download and decoding for the topo map widget of the editor (editorGeoMap). The place clicked becomes the geographic reference of the new scenery (//$g PUWG1992), and the scenery starts with the orthophoto of the editor on (//$o), lying under it in place. Not built on its own: the tree builds again with the selftests commit, the one before the translation. * feat(editor): the tools wired into the editor mode and the build editor_mode takes the tools of the commits before: its header declares them, editormode.cpp and the track mode drive them, and the renderers draw their overlays (opengl33 and legacy). CMakeLists lists the new sources and adds WITH_EDITOR_SELFTESTS (off by default); the unit tests build the new test files. Not built on its own: the tree builds again with the selftests commit, the one before the translation. * feat(editor): the selftests of the editor split by what they test The scripts of EU07_EDITOR_SELFTEST run the commands of editormodeselftest*.cpp: scene, track, rails, join, switch, signal, sleepers, sweep, terrain, monkey. The code is in every build; only with WITH_EDITOR_SELFTESTS=ON does the editor read the variables and run a script. The selftests of master now need the option too. The tree builds again with this commit. * feat(editor): the Polish texts of the editor, all of them in one update of editor_pl.po * fix(editor): the marks of the editor moved off the letters the starters read, and its lines told by their words The starters read some letters in a scenery file: the Starter the title //$n, the attachments //$f and the old scenery //$a, StarterNG and the launcher of the game the attachments //$f and the title //$n. The editor wrote its download area, shared grade lines and fitted tracks under them, and took every line of those marks as its own: saved, a scenery lost its title and attachments, and the starters showed the lines of the editor as them. - The download area is //$z, the shared grade lines //$w, the fitted tracks //$t. - Their lines and those of //$o are told by their first words, as //$h, //$r and //$k were: //$o only as "on 0" or "on 1", the origin or the settings of older builds after it, so a trainset described as "on ..." isn't taken. - The lines older builds wrote under //$a, //$n and //$f are read only by the words the editor wrote in them, or by its own comment line before them, under the new mark. Each save writes them again with it, also when the editor didn't change them; the lines of the starters stay. * fix(editor): a square of the terrain model cut short isn't read, and the squares are kept in the user's cache The service puts the directory of its GeoTIFF before the pixels, so a file cut short by a write or a fetch broken off was read whole, the pixels missing coming out as heights of 0 m, for good. - The reader (editorGeoTiff.hpp, with unit tests) wants all the strips the rows need, each within the file. - A square is written under a name of its own and renamed. - The squares are kept in the user's cache folder (nmt/grid1), as the orthophoto, since the game folder may not be writable; without one, in the game folder as before. The ones older builds kept there are moved over as each is asked for, a damaged one deleted. - Capabilities that didn't come aren't kept as the fallback name for the session: the next square asks again. * chore(editor): the http client said to serve the terrain model and the topo map too, the gif fix of stb_image marked local - CMake names the orthophoto, the terrain model and the topo map as what goes without libcurl. - The fix of the gif frame two back in stb_image is ours, not upstream's (v2.30 still has the old line): a comment at the line says so, for an update of stb_image to keep it. * fix(editor): the lines moved off the letters of the starters in a merged layer written once, not again with each save A merged layer's text goes into the file of its target on save, and the target is read again; the marked lines of the merged layer were dropped only for the marks the editor had changed. The lines moved to the new letters stayed with their old places, and the next save of the session wrote them into the target once more. - After a save, the marked lines of every merged layer are dropped; those of its target are read again from its file. - The lines of a dead layer aren't written again. - The selftest layermerge merges the layer of one file into another's, as the layers panel does. * feat(editor): the station as the signalling sees it, found from a switch of it, and its schematic editorStation: the switches joined by plain track no longer than 1.5 km, the tracks between them and the lines leading out, 400 m of each beyond the outermost switch; laid out as a schematic along the main track through the station. The track circuits, the signals and their lights, the lengths of the station tracks and the lines to the next stations are worked out from it for the signalling tool. Nothing uses it yet: the signalling tool, wired in the next commit, does. It builds on its own. * feat(editor): the signalling of a station in the Signalling tab, laid out from a switch of it as on the lines of PKP PLK LMB on a switch in the Signalling tab opens its station: the schematic in a window of its own, the name, the code, the line and the km of the station, and the steps of its signalling, each one step of undo: - the switches numbered along the station after its code, bpd_1, the parts of a slip bpd_15a..d; the drives and the objects naming a switch follow it, events written by hand keep the old names; - the tracks cut at the joints of the track circuits, Ie-4, and the circuits named, bpd_iz1, bpd_it1; - the entry, exit and shunting signals placed and named, A, F1, Tm1, from the templates chosen; - the signals of the lines out: the automatic block of three or four aspects, the block sections no shorter than the braking distance, or the semi-automatic one with a warning disc; block boundaries drawn by hand; - the lights each signal needs for its routes, Ie-1 and Ie-4 § 8, its template fitted to them; - the SHP track resonators the signals lack, rezonator.inc; - the lengths of the station tracks as a table, and what the station and its lines still lack, with Repair. The stations described are kept as //$s lines in the files of their layers. * fix(editor): the signals of a station and of its lines not placed with no template for a kind, no track cut for them first - Set the signals and Set the line signals cut the tracks first, and only then found no template for a kind. They stopped there, but the cuts stayed as a step of undo. Now they ask for the templates first, by the lookup signal_batch() uses (signal_template_for(), behind editor_signals::signal_templates_missing()). Without one nothing is cut. - Set the signals and Set the track circuits again under Track lengths had the ids of the buttons of the steps above them. ImGui flagged them when hovered. Each now has an id of its own. * fix(utilities): win1250_to_utf8 turns the polish letters into utf-8, not into themselves utilities.cpp starts with a byte order mark, so the compiler reads it as utf-8. The game's targets don't get /utf-8, though: set_target_properties(... COMPILE_OPTIONS ...) in the MSVC block of CMakeLists.txt replaces the options add_compile_options gave them. So the string literals went into the program in the code page of the system, windows-1250 on a polish one, and the map of win1250_to_utf8 turned each letter into itself. What read windows-1250 through it got no utf-8: the names, descriptions and categories of the sceneries and the names of the textures in the launcher, and the descriptions of the templates in the editor. The editor writes such a description back through utf8_to_win1250, which makes a question mark of each byte that doesn't start a utf-8 sequence, so the letters it had read unchanged didn't come back. The utf-8 sequences are written as bytes now, the way utf8_to_win1250 has them, so they don't depend on the options. The options stay as they are: with /utf-8 other literals of the files with a byte order mark would change, the "±" of the table of Bezogonkow would take two bytes and move the letters after it. The unit tests check the letters and that the text given is valid utf-8. (cherry picked from commit |
||
|
|
8794c80c58 |
fix(models): a scenery light that is off no longer shows lit for a moment the first time it comes into view (#407)
* fix(models): a scenery light that is off no longer shows lit for a moment the first time it comes into view Since |
||
|
|
79c88c1793 |
Python screens: the worker's image goes in place only after its fence, not a bare glFlush (#411)
* fix(python): a screen uploaded by the worker goes into a texture of its own, put in place after a fence With python.threadedupload (the default) the worker wrote the image into the texture the drawing thread draws from, with glTexImage2D and glGenerateMipmap from its own context, and only a glFlush after. Nothing ordered the two contexts: the drawing one could sample the texture while its storage was being made anew, its level 0 half written or its mipmaps not made yet, and on a change of size the worker deleted the texture and wrote a new name into `id` with no synchronisation. The worker now uploads each image into a new texture, sets a fence after it and flushes. The drawing thread, once per frame in python_taskqueue::update(), waits for that fence on the gpu (glWaitSync, the cpu goes on), deletes the old texture and puts the new one in place. The worker no longer touches the texture drawn from, nor `id`. An image the drawing thread didn't take yet is dropped by the next one. Without sync objects the worker waits for the upload with glFinish. The windows of python.displaywindows with python.sharectx bound the texture once; they now bind the name of the last image put in place, which python_rt keeps for them. Uploads on the drawing thread (python.threadedupload no, the Lua and WebAssembly screens, the editor) are as they were. * fix(python): the image of the worker goes in place once its fence is signalled, the viewer binds it each frame From the review of |
||
|
|
1b26caca24 |
fix: inverted clamp bounds give the upper bound again, as the old clamp did (#412)
* fix: inverted clamp bounds give the upper bound again, as the old clamp did The project's own clamp() of 26.05 applied the lower bound first and the upper one last, so bounds given the wrong way round always gave the upper one. Its calls became std::clamp (undefined for such bounds) and, where the bounds can cross, safe_clamp, which swapped them instead. Neither gives the old result. - safe_clamp is the old clamp again, min(max(value, lo), hi). This restores 26.05 at every place it is used, among them the AI's dynamic brake test (its demand was fAccThreshold x1.21, the swap made it x1.01), the generator voltage, the heavy cargo acceleration of the AI and the EP correction of the brake. - The cab camera is kept inside the cab with safe_clamp. The bounds come from the .mmd as written and some cabs have them reversed (the B cars of SA134 and 21xM, ET40, 754); with std::clamp the camera jumped between the two bounds every frame on MSVC, as its result is fed back. * fix: every call of the old clamp gives what it gave up to 26.05.1 |
||
|
|
a1fbdf47c6 |
fix(camera): the cab shake taken at the frame's moment between its fixed steps, it no longer moves in steps (#403)
The shake of the cab camera (ShakeState) is stepped by update_shake at a fixed 1/50 s, while the camera read the raw state every frame. At any frame rate other than 50 the shake moved in steps: some frames showed the same state as the one before, others jumped one or two steps at once. Players with 144/165 Hz screens saw the cab jump around the camera. The vehicle now keeps the state from before the last step (ShakeStateBefore), and the camera takes the state between the two at the share of the step the simulation time has reached (shake_between). The offset and the velocity are blended, the angles worked out from the blended velocity. The shake itself is computed as before, with the same fixed step and the same springs; only what the camera shows of it changes. It lags the vehicle by a constant 1/50 s instead of 0 to 1/50 s. Measured with the same build and the blending switched off by a temporary environment variable (not in this commit), the EU07 on td.scn driven by the AI, frames over 30 km/h, two runs of each: - 60 fps: frames with the shake standing still 25% -> 0%, the change in its speed frame to frame 4-5 times smaller for the offset, a third smaller for the angles - ~144 fps: 65% -> 0%, 10-18 times smaller for the offset, 4 times for the angles - no fps limit (~520 fps): 90% -> 0% Co-authored-by: maj00r <maj00r@host> |
||
|
|
72c69d9249 |
Fix stutter with fpslimit on Windows and a texture unit cache race with threaded Python screens (#399)
* fix(render): the python upload thread only marks the texture unit cache stale, the drawing thread clears it
update_from_memory() runs on the python upload thread with python.threadedupload (the default), and since
|
||
|
|
4b80d3c428 |
fix(particles): the smoke of a cold diesel keeps the colour of its template (#413)
|
||
|
|
8bda0275f3 |
fix(camera): the roll of the cab shake tilts the view about the line of sight again (#414)
|
||
|
|
267d5b3a8f |
AI: shunting brakes for a stop point again; test mode: no false SPAD for a train set across a signal (#409)
* fix(ai): shunting AI brakes for a stop point again, the braking curve doesn't plan from a brake table left at zero The braking curve outside the final approach plans from the consist brake table since |
||
| 80fd5aad2b |
Merge pull request #385 from MaSzyna-EU07/feat/blackbox-standalone
feat(blackbox): the black box on its own, lighter physics logs, physicslog.trains |
|||
| 388b8407a5 |
Merge pull request #386 from MaSzyna-EU07/fix/editor-trackbed-texture-crash
fix(editor): the texture pickers let go of the images scrolled past, the memory no longer runs out |
|||
|
|
bcecfe2964 |
refactor(renderer): a material's textures let go of in one place, for every backend
Unload_Material walked the textures of a material and of its variants the same way in both opengl renderers, and Sonar counted its declarations, set into the long run of overrides the two headers share, as duplicated: 5.2% of the new code, 3% allowed. The walk is gfx_renderer's now, over IMaterial::GetAllTextures and the backend's Unload_Texture. Materials that don't tell their textures, the Vulkan renderer's, give none, and are left alone as before. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>build-fix-editor-trackbed-texture-crash |
||
|
|
16cad1fc07 |
Merge pull request #332 from MaSzyna-EU07/fix/editor-po-wizard-newlines
Editor: the Polish of the download area tooltip and hint shows again |
||
|
|
a0d7b86d03 |
Merge pull request #362 from MaSzyna-EU07/fix/linux-file-case
fix(files): data named in another case is found on Linux, folders included |
||
|
|
c6e7315bc1 |
Merge pull request #380 from MaSzyna-EU07/fix/editor-ca-bundle
fix(editor): the certificates of https looked for where the distribution keeps them |
||
|
|
ba4bea92e8 |
Merge pull request #381 from MaSzyna-EU07/feature/no-avx-build
ci: Windows and Linux builds without AVX2, named _noavx2, in the public download |
||
|
|
c745909488 |
Merge pull request #383 from MaSzyna-EU07/fix/legacy-ui-stale-texcoord
fix(renderer): the legacy renderer clears the arrays the scene left before it draws the ui |
||
|
|
b48694ba3c |
Merge pull request #384 from MaSzyna-EU07/fix/unknown-renderer-falls-back-to-full
fix(app): a renderer this build doesn't have starts the full one, not the legacy one |
||
|
|
0a1c4fd425 |
fix(physicslog): what no file name may have on Windows is replaced in the name of a log, on every system
A spawned vehicle is named <spawn>:<number>, and its log went to physicslog/<spawn>:<number>.dat. On NTFS that isn't even an error: it is the stream <number>.dat of the file <spawn>, which nothing lists, so the log was written and no one could find it. On Linux the file was there, under a name Windows can't open. The name of the file of a log has < > : " / \ | ? * and the control characters replaced by _, on every system, so the logs are named alike everywhere (spawn_ezt:12 is spawn_ezt_12.dat), and a _ before a name Windows keeps for a device (con, nul, com1..., with an extension too). The bytes of 1250 stay. The black box tells the log it begins by the same name. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>build-feat-blackbox-standalone |
||
|
|
5498995672 |
fix(blackbox): the tapes go on after the editor, with the physics logs off
The tapes of the user's vehicles were written by update() after the check for the writer of the recording, which close() drops on entering the editor (F11). With "physicslog no", the speed recorder's tape ended at F11 while the drivers' own logs of "physicslog yes" went on. The tapes are files of their own: they are kept from the moment the recording was to be made, the file made or not, closed or not, until the simulator closes, after the gap of the time in the editor. Only the recording itself stops at F11, as before. Reported in the review of #385. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
|
99b93abd64 |
fix(physicslog): nothing that goes wrong in the writing ends the simulator; a name in 1250 makes a path on a Windows set to UTF-8
The writer's thread called flush() with nothing around it, so an exception leaving write() ended the process with std::terminate, with no message. One that happens: the file of a log was made a std::filesystem::path of the vehicle's name, which MSVC converts in the code page of the system and throws for bytes not of it; on a Windows with "Use Unicode UTF-8 for worldwide language support" a vehicle named in 1250, as the sceneries have them (0xB3 for ł), isn't UTF-8. - the file of a log is made of the name with MultiByteToWideChar without MB_ERR_INVALID_CHARS: the same file as before for every name of the code page, U+FFFD for the bytes not of it instead of a throw. Elsewhere the bytes are the name of the file as they were - a log whose writing throws for any other reason (memory) is kept for the next try as one whose file can't be written, and the other logs are written still - the thread, flush_for() and stop() write through flush_safely(), which lets nothing out: what was taken for a writing failing as a whole is lost and told in the log, the next writing goes on - the messages name the file by the bytes of the name, not by path::string(), which converts back Tests: a log which can't be written doesn't keep the others from theirs; a vehicle named in 1250 has its log, read back by physicslog_writer::file(). Reported in the review of #385. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
|
b832d2fab9 |
fix(editor): the image of a material with no texture in its first slot is let go of too
Its normal and specular maps were read with it and stayed in the memory: the trim passed over every image with no texture of its own, the material along with it. |
||
|
|
423d49b42c | fix(physicslog): every stop asked wakes the writer, the one from itself too; no psyche for a driver of no vehicle | ||
|
|
0bbf07bf11 |
fix(editor): the images of the materials let go of with all their textures, the browser's Refresh a frame later
An image of a material read the material with all its textures, the normal and specular maps and the variants, and only the first one was let go of: those stayed in the memory for every material scrolled past. They are let go of together now (Unload_Material). texture_manager::unload also takes the textures read and not made yet, waiting for a worker still reading one, and deletes the gl texture itself: release() would read it back from the card with gfx.resource.move only for it to be dropped. The Refresh of the texture browser let go of the images in the middle of the frame, while panels drawn before it could still refer to them; it now marks them, and they go at the start of the next frame. Checked on td.scn by a scripted run: 8040 images of the picker read a batch every few frames end at 2.0 GB, against 2.3-2.6 GB before this and 1.1 GB at the start; a trackbed material picked from the folder is drawn on the track after its image was let go. |
||
|
|
94b7c7f01b | fix(physicslog): the writer's destructor throws nothing, its wait a plain condition variable woken to stop; a driver of no vehicle controls nothing | ||
|
|
4525023777 |
fix(editor): the texture pickers let go of the images scrolled past, the memory no longer runs out
The picker of the textures of a track offers every texture of the textures folder, and each tile that came into sight read its texture full size, into the graphics card and the memory, and kept it there for good. Scrolling through the list for a trackbed texture took about 2.5 MB a tile: some 900 tiles made 3.5 GB, the whole folder 28 GB, and the game went down once the memory ran out. An image out of sight for 3 seconds is now taken out of both again (texture_manager::unload, Unload_Texture of the renderers) and read anew when it's shown once more, a few a frame as before. Only a texture the scene has never drawn is let go; those of the scene stay to the resource sweep. The texture browser lets go of its images on Refresh too. The picker no longer lists textures/previews: the 4100 previews of the node bank are images of models, not textures of anything. Checked on td.scn by a scripted run: 8000 images of the picker read one batch a few frames after another held the game at 2-3 GB, where it went to 28 GB and fell before; images shown again after they were let go come back; a trackbed texture picked from the folder is drawn on the track after its image was let go. |
||
|
|
b421550cbe |
fix(physicslog): a train's log begun at its first line, so passive crews get no file
Every crew is primary in the constructor and the drivers but one of a train with crews in more vehicles are made passive only before it starts, so a log begun there left files with only the header. With physicslog on all logs begin at once as before; one of a train waits for update_logs. |
||
|
|
b7950af6b6 |
feat(physicslog): the logs of the trains only, physicslog.trains
"physicslog.trains yes" in eu07.ini writes the .dat log of each driver who drives rail vehicles, not of the passengers nor of the road vehicles that make "physicslog yes" hundreds of files on a big scenery: the tape of every train. A driver made primary later begins its log then, and isn't counted while it doesn't log. The black box goes with it when the ini says nothing of its own, and doesn't write a second log of a driver writing its own. The doc tells the doors' sides are the vehicle's, the driver's swapped in the rear cab. |
||
|
|
043b9fb2f0 |
fix(physicslog): failed logs kept within 64 MB in all, cut outside the lock, never written twice; the writer knows itself when stopped
The lines kept while files can't be written are cut to their share of 64 MB before the lock the simulation waits on. A log written in part is cut back to its length, so the next try doesn't write its start twice. A stop from the writer itself is told by its id even after another thread took it to join, and the watchdog's flush waits at most its time, the writing on a thread of its own. |
||
|
|
5d0bf6ed00 |
fix(blackbox): the wheels of the powered vehicle, as the speedometer has them, and the logs described as the drivers' they are
In a unit driven from an unpowered car, the speed of the wheels was of the car the user is in, whose wheels never slip; now of the motor car, signed as the occupied vehicle's speed. The doc tells of a log per driver, and of the two seconds of real time a crash loses. |
||
|
|
7547e6bf07 |
fix(physicslog): a log whose file can't be written keeps its lines and header, and the writer is stopped once on every way out
A failed write put its lines back, ahead of those since, so a file held by another program for a while gets its header and every line once it can be written; up to 4 MB kept per log. Stopping is joined once, and not from the writer's own thread. main stops the logs after an error it caught too, and the test watchdog flushes with a time limit, so a disk that doesn't answer doesn't keep the run from ending. |
||
|
|
aaff0a430b |
fix(settings): the gfxrenderer word read in any case, as the starter reads it
A hand-written "gfxrenderer Legacy" started the legacy renderer only because every unknown word did; with unknown words going to the full renderer it would have changed. The word is lowercased when read, as StarterNG's RenderEngineOf does, so Legacy, Simple or Full mean what they say.build-fix-unknown-renderer-falls-back-to-full |
||
|
|
68c33922a6 |
fix(app): a renderer this build doesn't have starts the full one, not the legacy one
The starter writes "gfxrenderer vulkan" for the Vulkan renderer of the builds that have it. This exe doesn't know the word, and every word other than default/experimental fell to the legacy renderer: a player who picked Vulkan got the slowest renderer, the one whose ui crashes on NVIDIA (the arrays left by the scene). "experimental" in a build without the NVRHI renderer ended with "no renderer found!" and no game at all. After the settings are read, a gfxrenderer word this build has no renderer for becomes "default", with a line in the log saying so. The words it knows stay as they were: full/default, legacy, simple, null, the older starters' simpleshader and off (legacy as before), and experimental where it is built. |
||
|
|
88522fe600 |
fix(renderer): the legacy renderer clears the arrays the scene left before it draws the ui
The legacy renderer drew the ui with the client arrays of the last scene draw still enabled: the texture coordinates of the normal map unit and the normals, pointing into the vertex buffer of whatever was drawn last, 48 bytes a vertex. The ImGui backend sets up the vertex, colour and texture coordinates of the active unit only, so the driver read the stale arrays with the indices of the ui. Once the ui had more vertices than that buffer, the NVIDIA driver read past its end and the game crashed; a crash dump of the editor (an area selected, a track being laid) shows the read of vertex 15113 at a 48-byte stride in ImGui_ImplOpenGL2_RenderDrawData. The arrays are now disabled on every texture unit the geometry uses, with the normal and colour arrays, inside the glPushClientAttrib already there, so the scene gets its state back after the ui. The default and the experimental renderers don't go through this code.build-fix-legacy-ui-stale-texcoord |
||
|
|
0916104293 |
ci: Linux is built without AVX2 too, published as maszyna-linux-x64-noavx2.tar.gz
The Linux job becomes the same two-variant matrix as Windows (ENABLE_AVX2=ON/OFF), each variant with a concurrency group of its own. The publish job downloads both Linux artifacts, restores the executable bits in both and releases maszyna-linux-x64.tar.gz (AVX2, the old name) next to maszyna-linux-x64-noavx2.tar.gz. The no-AVX2 executable is eu07_<date>_<hash>_noavx2.build-feature-no-avx-build |
||
|
|
507fcd8cb3 |
build: a build without AVX2 has _noavx2 at the end of its executable name
With ENABLE_AVX2=OFF the executable is eu07-x64_<date>_<hash>_noavx2.exe (eu07-x64_noavx2.exe without ADD_VERSION_SUFFIX), so a starter can tell the two builds apart by the name alone and pick this one on a CPU without AVX2. The AVX2 build keeps its name. |
||
|
|
348a814ce2 |
fix(ci): the no-AVX2 Windows build has its own concurrency group and its own public download
After master was merged the two Windows variants shared one concurrency group with cancel-in-progress, so one cancelled the other, and the publish job looked for an artifact name the matrix no longer produces. Each variant now has its own group, the publish job downloads both, and the release carries maszyna-windows-x64.zip (AVX2, the old name, so the links already handed out still work) and maszyna-windows-x64-noavx2.zip. The AVX2 variant passes -DENABLE_AVX2=ON explicitly instead of an empty flag. |
||
|
|
6ffbc0a6f6 | Merge master into feature/no-avx-build | ||
|
|
465300df8d |
fix(editor): an https proxy checked with the same certificates
CURLOPT_PROXY_CAINFO gets the file found for CURLOPT_CAINFO, so a download through an https:// proxy on Fedora or RHEL doesn't fail on the proxy's certificate.build-fix-editor-ca-bundle |
||
|
|
4558976733 |
fix(editor): the certificates of https looked for where the distribution keeps them
The Linux game bundles the libcurl of Ubuntu, which knows only Ubuntu's /etc/ssl/certs/ca-certificates.crt. Fedora, RHEL and openSUSE keep the certificates elsewhere, so every download of the editor (orthophoto, terrain model, topo map) failed there with a certificate error. The file is taken from SSL_CERT_FILE or CURL_CA_BUNDLE when set, else the first of the files the distributions keep, and passed as CURLOPT_CAINFO. With none found libcurl keeps its own. |
||
|
|
e413675c02 |
Merge pull request #369 from MaSzyna-EU07/fix/ci-openal-audio-backends
fix(ci): build OpenAL Soft with the PipeWire, PulseAudio and ALSA bac… |
||
|
|
73d535f73e |
fix(ci): build OpenAL Soft with the PipeWire, PulseAudio and ALSA backends, so the Linux game has sound
The Linux job builds OpenAL Soft 1.24.2 from source, but the runner had none of the headers of its audio backends, and OpenAL Soft's CMake leaves out every backend it can't find. The bundled libopenal.so.1 had only OSS, Null, Wave and Loopback. Modern distributions and the Steam Linux Runtime have no /dev/dsp, so the game logged "Failed to obtain audio device" and played nothing, with any Steam Linux Runtime and without one. - install libasound2-dev, libpulse-dev and libpipewire-0.3-dev before building OpenAL Soft - configure it with ALSOFT_REQUIRE_PIPEWIRE/PULSEAUDIO/ALSA, so a missing header fails the build instead of shipping a library without sound The backends are loaded with dlopen at run time (ALSOFT_DLOPEN is on by default), so the bundled library gains no new dependency. Tested by putting an OpenAL Soft 1.24.2 with these backends next to the game from the Steam depot: under Steam Linux Runtime 4.0 the sound works. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>build-fix-ci-openal-audio-backends |
||
|
|
7b7529e24e |
feat(blackbox): the cab record holds the speed of the wheels, apart from the vehicle's when they slip
The speed of the cab record is the vehicle's. The speedometer, and a speed recorder's tape with it, shows the speed of the wheels as they turn, which runs away from it while they slip and drops to 0 while they are locked in a skid. It is added at the end of the record (194 bytes), where the readers skip what they don't know; the .dat tapes have it already, in their second column. |
||
|
|
03a3b4c09a |
feat(physicslog): the logs written every two seconds by a thread of their own, no file kept open; the black box's tapes as it goes
"physicslog yes" kept the .dat of every vehicle with a driver open for the whole run and flushed it twenty times a second from the simulation's thread: some 580 open files on Tomaszewo. The lines are kept now and a thread of their own writes them every two seconds, each file opened only for the moment it is written to. The files, their lines and their text mode stay as they were. - The black box, with the physics logs off, writes the tapes of the user's vehicles through the same writer as it goes, not at the end. A log begins when the user is first in the vehicle and goes on while the user looks around outside, until another vehicle is taken; back in it, the log goes on after the gap. The vehicle's own driver keeps the log only if it is the primary one, as a passive one's clock stands still; otherwise the driver of its train does. - What is kept is written when the simulator closes, after the black box, at the end of a test run, and when the test watchdog ends a run. A crash loses the last two seconds. - On Tomaszewo with the black box alone, the one tape grows every two seconds. With "physicslog yes", the 581 logs grow and the simulator holds as many handles as without them; writing them takes the thread about 1.9 s, which the closing waits for once. |
||
|
|
5ca2f7e38f |
feat(blackbox): the black box on its own, writing the tapes of the user's vehicles at the end
"blackbox yes" in eu07.ini records the black box with the physics logs on or off, "blackbox no" never; with no entry it goes with "physicslog yes" as before. With the physics logs off, the recorder keeps the line of the physics log of the driver the user is with, every 0.05 s, and writes it to physicslog/<vehicle>.dat when the recording closes: the speed recorder's tape of the vehicles the user was in, without every vehicle keeping a log open and written twenty times a second. The line and the header of the log are made by one function of the driver, which its own log writes too, and the file is opened the same way, so the two are the same to the byte: the readers of these logs (the old Hasler tape viewer, the starter) read them as before. |
||
|
|
0488519f39 |
fix: the simulator opens more than 512 files at once, and a shader it can't open says why
With "physicslog yes" every driver keeps its .dat open, and a large scenery (Tomaszewo: some 500 vehicles, cars of the roads included) took all 512 streams the MSVC runtime gives: the next fopen failed and the simulator stopped at "cannot read shader: shaders/map.vert" on opening the map. The limit is raised to 8192 at the start (the soft limit of open files up to the hard one on Linux), and a shader file that doesn't open is told as such, with the system's reason, rather than as one that can't be read. |
||
|
|
38540d08f5 |
refactor(files): named functions instead of the lambdas of the folder cache (Sonar S2259, S3608, S6004)
Sonar took the folder captured by the lambda reading it for a reference to null, failing the quality gate on reliability. The folder is read and the path kept by functions in the anonymous namespace, the age of the listing declared in its if.build-fix-linux-file-case |
||
|
|
d76e6368b0 |
fix(files): models named in another case are found on Linux
The manager of the models looked for the file with FileExists of a single name, a plain check again since the banks keep the names asked for, so a model named in another case (models/przejazdy/plyty3_L, dynamic/pkp/st44_v2/main/2M62-0685-A) was reported missing before TModel3d::LoadFromFile() could open it as it's named on the disk. find_on_disk() finds the file in any case and gives back the name asked for, the key of the bank. |
||
|
|
f9b7e9174c |
fix(files): terrain files named in another case are found on Linux
The terrain directive names .txtf/.btf files in lowercase, as the parser gives every token. The name is now taken as it is on disk, from the text file first, so the binary file is made next to the text and the editor's references and layers keep one name. |