- User Since
- Jan 23 2017, 5:13 PM (297 w, 3 d)
Aug 4 2022
Mar 9 2022
Feb 23 2022
Feb 22 2022
Jan 15 2022
Restore back to the state before debugging macOS failure.
Ah, I see what's happening.
OK: I found the following bug filed on mozilla's tracker: https://bugzilla.mozilla.org/show_bug.cgi?id=1635481 which has almost the exact same stack trace. That was affecting a user on macOS 10.14 with python 3.7.
Another pass on the macOS CI : Try reverting a mozilla change from two years ago, see if it does anything.
Get the macOS CI to rebuild spidermonkey so I have a comparison build log to compare to.
Oct 22 2021
Sep 25 2021
Rebase now that parent revision has been committed
Sep 23 2021
Intersect camera at corners of polygon, instead of using an imaginary horizontal plane based on averaged terrain height (see previous diff for that).
Attempt to use the average height of terrain in the camera's view, instead of just the height of the centre point of the view.
Sep 22 2021
Use const, and reduce number of method calls on the g_Renderer object.
One (final) pass on the CI
Sep 21 2021
Sep 18 2021
Sep 6 2021
Make the define CONFIG2_VALGRIND, as suggested by @sera
Jul 1 2021
I can't see how it would: we're not tagging anything to be translated here, just using translations that already exist.
Jun 30 2021
Jun 26 2021
Jun 10 2021
- Rejig CInput's text-handling to prevent infinite loop when using IGUIObject::getComputedSize() in CGUIScrollBarVertical::Setup()
- (See full commit message starting "Add a flag that's set [...]" under "Revision Contents" > "Commits" below)
- Use IGUIObject::getComputedSize() in CGUIScrollBarVertical::Setup()
- Calculate content size on demand, instead of reading from a cached value.
Jun 9 2021
(Missed a file.)
Debundle and make Valgrind optional
@Angen, you'll still get error messages, however:
- There should be fewer error messages emitted, as we now handle things being undefined better;
- Error messages should be issued less often, as follows:
- If it's a civ-specific template that's broken (e.g. structures/brit/arsenal) you should now only get error messages for that once (and not every time the template is requested somewhere);
- Ditto for broken technologies and auras;
- If it's a parent template that's broken (e.g. template_structure_military_arsenal) you'll get error messages for each civs' specific version of it (as you navigate down the "Civilizations" dropdown), but only the first time a given civ is selected.
- Error messages concerning broken Entity templates should be reduced, as you no longer get a traceback of the JS code (which isn't particularly helpful in this case).
- (Technology and Aura breakages still get a traceback, but that's part of the behaviour of the JSON parser on the c++ side.)
Also handle broken technologies and auras
Jun 7 2021
With the suggested changes below applied on an ArchLinux system, userreport_hwdetect.txt reports:
Jun 5 2021
May 25 2021
May 23 2021
- Rebase on top of recent changes
- Some changes in response to @Stan's comments
- Fix bug introduced in earlier version that caused click-dragging the scrollbar's bar to not work correctly
- Draw scrollbars on top of the background sprite of the gui elements
May 22 2021
Revision now contains just the changes for this revision, and not also the changes of its parent.
May 16 2021
I wonder... do we have any players from the Indian Subcontinent? (e.g. India, Pakistan, Bangladesh, Myanmar, Sri Lanka)
May 9 2021
A version of this that doesn't move the Win32 build artefacts.
Yeah, it was a typo that sneaked in (should be miniupnpc not minipupnpc - too many ps). I've fixed it locally, but its not really worth updating the revision/running the CIs for a single character change.
May 8 2021
May 3 2021
Upload without CI-only debugging commit
Apr 22 2021
Yeah... the JS environment has no way to tell if a file has been modified since it was last requested from the VFS. And considering that the VFS has its own caches - and hot-loading of simulation data files is not implemented - even when bypassing the JS-side cache the old file would still be loaded from the C++-side cache.
Apr 17 2021
Apr 14 2021
Mar 26 2021
Satisfy linter (hopefully)
Mar 25 2021
Rewrite the template base/variant detection code, and how its used.
Mar 23 2021
@Angen, I think I've come up with an approach to this I prefer: may I commandeer this?
Mar 7 2021
Mar 6 2021
Change cp arguments (the MacOS version of cp lacks args that the GNU one possesses.)
Use cp instead of ln;
Mar 3 2021
Change mind about miniupnpc; realise that zlib does support pkg-config
Attempting to work out why isn't it able to find fmt.pc...
Make sure we have the .pc files