alpha wrote:Alright, so to fix this problem, we have 3 options:

1. "Download server list" button aka warsow style
2. Download server list on "multiplayer" window click ([-z-]'s idea)
3. Manually mess with autoexec config and play with console on every game launch

How about we don't enforce a use of solution #3 ?

Solution 3 works fine.
Just make another nexuiz starup script the executes a cfg file that has the addional option and then executes your normal stuff.

That button would be an improvement for 1%, and slow 99% down. No.

divVerent wrote:That button would be an improvement for 1%, and slow 99% down. No.

Slow down how much 1 button click?

alpha wrote:

divVerent wrote:That button would be an improvement for 1%, and slow 99% down. No.

Slow down how much 1 button click?

I've played some games with the update button rather than nexuiz's automatic list. They feel much less polished. You can't just jump into a game, you have to press the damned button first before anything. With the automatic list the list populates while maybe you are editing your player info, your settings, etc.

I agree, warsow pisses me off for this very reason.

[-z-] wrote:I agree, warsow pisses me off for this very reason.

Does every other game that does not connect to internet at start piss you off? (half life ?)

Heck, whatever you say, the current method of querying for servers on startup will stay. It is user friendly, and fits well with the game, and saves lots of players about 5 to 10 seconds every time they start the game.

As stated before, the real way to solve this would be to use a thread for querying the server list, at least to get timeout functionality.

I can implement that for Linux and OS X, but someone else would have to do it for Windows.

As simple as that...

I have working server timeout code here:

http://emptyset.endoftheinternet.org/~r ... meout.diff

The Windows part got tested only on wine.

This limits the waiting time to 2 seconds per DNS resolve, that is, a total of 8 seconds on Nexuiz startup.

I can't commit it without clearance from LH though.

well NOW we're getting somewhere

indeed

Thank Div0 for this. Really this is amazing.

Anyone tried it?

divVerent wrote:Anyone tried it?

It works like a charm

You should use _WIN32 for MSVC.

I haven't seen the hardcoded Master Server IP, but fallback might be already there, so sorry for bothering if that's the case.

We don't use MSVC, and I just used the same define also used in other places.

I mean you (darkplaces) still provide MSVC project files (not that you use it) and MSVC uses _WIN32. WIN32 is defined internally by some windows header files and those aren't included one needs to use /DWIN32 to compile. Same can be said about NetRadiant.

I have no idea about this, and won't dare to change this as this MAY break compilation with mingw, or even cause subtle runtime issues.

I would however take a patch changing this from someone who actually TESTED it (but then, with mingw too!).

BTW, NetRadiant is ONLY compiled with mingw, and has no MSVC support. But, feel free to create project files for it if you want.

http://lists.boost.org/Archives/boost/2004/07/68241.php
http://predef.sourceforge.net/preos.html#sec24

You can use (WIN32) || (_WIN32) if cautious. WIN32 is defined in windef.h

I am interested in compiling darkplaces with intel c++ compiler, which is free on linux to further optimize dp for atom on netbooks.

Anyway, all existing DP code uses the WIN32 macro, and I'm not going to change this without LH's okay, as the existing code works fine. WHY it works fine, I don't care.

It works because it includes windows.h, which includes windef.h

I'm not asking to remove anything from dp, but you could use #ifdef (WIN32) || (_WIN32) in NetRadiant.

Against, as NetRadiant aims to be mostly code compatible with GtkRadiant, as long as there are other GtkRadiant forks out there that may have features NetRadiant could use.

If I did a batch change to change all these ifdefs, patches will be unlikely to still apply.

What I can do is add robustness by adding -DWIN32 to the compiler command line when building a Win32 build.

Actually, I just checked: the Makefile, which is an 1:1 port of the previous SCons based build system, always adds -DWIN32 when doing a Windows build. So in NetRadiant, this is just fine. Anyone adding a Visual Studio project has to read the Makefile anyway, as NetRadiant consists of lots of subprojects that need different compile settings (e.g. because there are multiple header files of the same name). I know it's ugly, but it matches what Id Software released as GtkRadiant and I'm not going to change that as long as there are other merge-worthy Radiant forks like DarkRadiant.

BTW, I just merged NetRadiant's q3map2 with VorteX's changes to the Urban Terror map compiler fork.

divVerent wrote:BTW, I just merged NetRadiant's q3map2 with VorteX's changes to the Urban Terror map compiler fork.

What new goodies does that bring ?

Should have I expected more?

Code: Select all

libs\math\vector.h(98) : fatal error C1001: An internal error has occurred in the compiler.
1>(compiler file 'msc1.cpp', line 1411)
1> To work around this problem, try simplifying or changing the program near the locations listed above.
1>Please choose the Technical Support command on the Visual C++
1> Help menu, or open the Technical Support help file for more information

Btw,

Code: Select all

#ifndef boolean
#ifdef __cplusplus
#define   boolean   bool
#else
#define boolean int
#endif
#endif

should be after

Code: Select all

#ifdef _WIN32
#include "windows.h"
#include "stdio.h"
#endif

in radiant_jpeglib.h otherwise it overrides internal windows header boolean definitions.

That file is unmodified from idsoftware's

And what does it override? How does windows.h define boolean?

If windows.h does define boolean, I rather suggest that this block should be skipped instead. So I moved the windows.h include more up. Please test.

It's rpcndr.h which:

Code: Select all

typedef unsigned char boolean;

Then we get this error: http://msdn.microsoft.com/en-us/library/ws579tt7(VS.80).aspx

IdSoftware used(-s) STLPort instead.

STLPort instead of what?

Why can't you just use STLPort too?

They use stlport in zeroradiant (it was required last time i compiled it) instead of relying of correct microsoft stl implementation.

Tested your commit. jpeg6 now builds correctly.