Sorry, I never got around to writing proper docs for fteqcc. Someone else beat me to it, so there never seemed much point. Admittedly those prior docs are woefully out of date now, but hey...
Basically, C syntax is meant to work, unless it conflicts with existing QC syntax.
The options dialog of fteqcc should document the basic idea of most compiler flags.
If you set pr_checkextensions to 0, then AD will no longer be able to detect any QC extensions at all, and will thus use its entity-based particles instead (the exact same behaviour as in regular Quakespasm).
(Unfortunately AD has no way to select between entity particles and engine particles - if it thinks that it can use engine particles, it WILL use them.)
IIRC, the other option is to set temp1 to 1024, which will disable AD's engine+entity particles alike.
Using the provided updated fteqccgui.exe, anytime you edit a .qc file within the program it causes EOF errors.
Add a variable that isn't used anywhere into a def file.
Double click on the warning for no references.
In the subwindow that pops up, comment out a variable.
Witness the ERROR EOF detected in comment of file.
Using the provided fteqccgui.exe (ONLY one with support for compiling the csqc files which is why I'm considering this a bug report related to QSS), I get multiple erroneous no reference warnings for variables. After commenting out the related variable, I get numerous errors for not having defined that variable.
'EOF inside comment' means that there as a multi-line comment that wasn't terminated.
//single line comment
/*unterminated multiline comment that will hit eof
otherwise I've no idea what you're actually doing, and thus no idea how to reproduce.
Regarding unreferenced variable warnings, I'm going to need a proper example/more info.
If you're using '#pragma sourcefile' then remember that its compiling your code twice, and variables that are unused in your csqc are likely NOT unused in your ssqc (and vice versa).
If you're lazy, you can use the __unused modifier to just mute the warnings, but I wouldn't recommend doing so with fields.
Eof: Simply using // to comment out variable definitions e.g.:
// .float gorging;
Which then gives me that error for EOF on that line, but only if I add // inside FTEQCC's gui editor. If I add // in Notepad it's fine.
Unreferenced warnings (about 133 for my project) are showing that a variable is not used in this new version while the older version of fteqccgui has 0 warnings. The "unused" variables are clearly used by the code so some check must be either not parsing all files or failing to parse all files to see if they are used.
Tested without csqc and no #pragmas.
the 'EOF inside comment' error message is specific to multiline comments, and I have absolutely no explanation that would not be accompanied with a 'file contains null bytes' warning.
All I can really say is that there's more going on here than it seems, and I've no idea how to reproduce it.
The second bug is quite probably the same root cause as the first.
If editing your code works in notepad then you may have to just stick with notepad. You can get a commandline-only version of fteqcc here: http://fte.triptohell.info/moodles/fteqcc/
There's also some docs there for fteqcc's extra syntax, if anyone needs it.
I'll have to play with it more later and try to narrow it down. Thanks though.
what does it mean the console output
disabling rendering/network isolation
Its a slight rework to host_maxfps.
Its enabled when above host_maxfps is above 72(or 0), and ensures that c2s packets as well as the server itself are throttled to no more than 72 fps without throttling the client too.
This means you can set host_maxfps to whatever you want and no longer get any physics issues because of it. Like all the other engines that already fixed it...
I should probably just remove the prints, at least once I'm sure it all works properly. I don't really expect any issues, I'm just paranoid.
btw. for some reason i got this rendering borkage (the grey screen)while i'm running the frogbot mod with qss
You should advertise that in the main post or something.
-Attention all those who own a fancy-schmancy high fps monitor! QSS/FTE now have a fix for high framerate physics bugs! Rejoice-
I would test in glee, except my poor-man monitors are stuck in the 90's.
High Fps Normal Physics
If it works flawlessly, would definitely join the tier 1 level of enhancements that improved the Quake experience.
It is a great thing that Spike has become more and more immersed in the world of traditional single player Quake.
(And this implementation is the least complicated one I've seen. I saw something about "bf" timing. Ironically, mh's implementation for allowing high fps normal physics had lots of comments addressing bonus flashes, I can't recall if he addressed that -- I think he did and it required an extra clock at a minimum.)
/I read the code 2-3 hours ago because I had a "WTF??" moment when reading that.
I tested it a little bit last week on a 165Hz monitor and it looked/felt amazing. Agreed, this is a big deal! Check it out of you have a >60Hz monitor. (host_maxfps 0 = uncapped.)
The particle production rate was visibly a bit off @165Hz, but this is a well know issue discussed on insideqc.
I ran it through a few input scenarios that I thought might be able to "mess it up" but not seeing any juttering with some stuff that used to very subtley trip up DirectQ's implementation.
Also did a quick test connecting to an online server to see how it behaved.
Everything looked and felt completely normal and fine.
Regarding Future Implementation In QS Vanilla
my IRC update code gets called from the _host_frame code in host.c
Is it likely that I'm going to have to significantly change the way I'm calling this when the physics and framerate are separated?
A friend of mine would like to run Quake on 144hz monitor, but with 72fps physics. Using QSS by default the game run @ 144fps, and that does changes physics a lot. How to set original Quake physics while having 144fps ?
@spike @Qmaster Re: Code Parsing Fuckups
Hm,so the code works when written in Notepad but not in the actual editor? I had the exact opposite thing happen to me recently when trying to add entity information to a config file in a Doom mapping program I use, so I wonder, was this program originally written for Unix compliant OSes (Linux Mac etc) or Windows? The text parser might be expecting DOS-format end of line characters and not getting them, either that or there's a bug in the code that is searching for the end of comment delimiter, I haven't looked at the code yet so I'm not sure which is the case but that is probably where I would check first. Then again I am a pretty inexperienced coder so maybe you shouldn't take my advice 😂😂
The specific patch is quite small, and if you're doing something that's sensitive to framerates that that would exist regardless of the rate at which the server runs at. So no, just the usual potential-but-trivial merge conflicts.
QSS ALWAYS limits its physics to 72fps or less, regardless of video framerates, so its nothing to worry about with the latest QSS.
Just set host_maxfps to 0 and enable vsync and you're fine (usually - nvidia drivers seem to have issues with vsync and otherwise-high framerates).
If physics feels different then it'll be purely down to the interpolation (or if you're more familiar with higher rates - which are generally considered buggy).
I didn't add any settings to limit physics to eg 20fps as is the default on many dedicated servers. my aim was purely to smooth out the single player experience, and I personally hate 20fps anyway.
fteqcc accepts windows, unix, and supposedly also mac line endings, which is meant to match scintilla's behaviour also.
note that support for mac endings means that rogue's sourcecode does not compile as-is due to a mac line ending appearing in the middle of a single-line comment, and these being line endings means the next line contains uncompilable gibberish.
this is why I've configured fteqcc's text editor parts (a third party widget called scintilla) to display explicit line ending chars if they appear to be mixed.
While fteqcc attempts to support various types of unicode encoding, it only considers the ascii chars as whitespace, so chars like non-breaking spaces that you might have pasted in will confuse the compiler but might be invisible in the editor.
Any non-ascii codepoints are pretty much ignored - they're accepted in names, and any non-ascii chars inside strings will be treated the same way as they always have - needing eg com_parseutf8 1 in order to display them properly in-game (instead of as multiple random red glyphs). As an english speaker that's also quite lazy, this hasn't got much real-world testing, just specific cases that required annoying copy+pasted glyphs that I can't even read, so it wouldn't surprise me if the unicode crap is completely unworkable as-is.
It'd be nice to have an actual copy of a bugged file so that I can figure out which chars are actually in there, otherwise I have absolutely no idea how to progress (and I'm too lazy/paranoid to try fiddling with my system locale etc to try to reproduce it).
Because I can't use the CSQC without using the new compiler version and I really like being able to edit within FTEQCCGUI for quick simple fixes so testing to help us fix this....
If I compile code that has a warning for a variable that isn't used anywhere, then // it out inside FTEQCCGUI, then compile, I get the warning about EOF in a comment on that line:
•If I then A close FTEQCCGUI, reopen and recompile, the file change is not saved (add popup warning to save edits?).
•Or If I then B manually save the file with // added within FTEQCCGUI from FTEQCCGUI's File->Save, close FTEQCCGUI, reopen, recompile, I no longer have the EOF warning.
If I open the same file within FTEQCCGUI and delete the // for that variable, then recompile, I get the EOF error again. I could have sworn the older version auto-saved any in-FTEQCCGUI edits before compiling, but it seems like something is screwy with that......
AHH!! THAT'S IT!
If I change the file within FTEQCCGUI, SAVE IT!!, then compile, then I get no errors. That's the bug, auto-save-all before compile is disabled. Can you please add that back or at least add a warning or prompt to "Save files before compiling?"
Still wierd how if I don't save before compiling, then compile, get error, then save, then compile, the error is still there.
World Size Limit
What is the proper way to compile big maps for quakespasm-spiked ? The default world size limit in Quake 1996 is 4096 units. Quakespasm/jackhammer is 8192 units (beyond that limit some bugs occurs in the game, even in QSS). If I go with higher limits (16384, 32768, 65536, 131072, 262144) ericw tools don't want to compile the map. Thanks
ter-shibboleth world is 65536 units,
not sure, but prolly you have to add -bsp2(or something) switch to your command line
@ericw Tools Says
Create the output BSP file in BSP2 format. Allows the creation of much larger and more complex maps than the original BSP 29 format).
Escaping The +- 4096 Bounds
gl_farclip 100000 (or whatever you need)
does bsp2 have anything to do with this, other than indirectly? (i.e. bigger maps - more leafs, clipnodes etc.)
BSP2 has nothing to do with the bounds.
Yeah Thought So.