You've broken downloads for all clients supporting FTE's chunked downloads.

O'rly? Cry me a river.

And what is broken, except download queue?

Btw, chunk size is 1024 byte, client sent 77 packets per second, that gives us 77*1024 bytes per second download rate maximum.
Standart download use 1350 byte chunk and same 77 packets per second, so even theoretical download rate of chunked download is WORSE than standart, I simply do not understand why this download type was implemented by Spike and where the advantages.

I supposed chunked download gives better download rate with HIGH ping because used not reliable data transfer, but it not, probably because I ported wrongly and still somewhere used reliable method even for chunked download. Have no time to check/fix this. But still even I fix this, chunked download will be worse, at least with low ping.

Yup, it's pretty annoying. Unfortunately some server admins blindly upgrade to the latest broken MVDSV release.

Even more unfortunate is it that FTE extensions are negotiated before the server version has been transferred, so blacklisting broken servers is no option :/


It doesn't work. Fire up FTE and try for yourself.


Well, you answer the question yourself just below, so I'll just leave it at that...


As soon as you have more than 1 frametime (that's 13ms for 77FPS), the normal download method just becomes worse and worse. At 26ms your download speed becomes half of that at 13ms, and 39ms is 1/3rd the speed etc. The advantage is quite obvious. And yes, it IS faster, it is with FTE, it is with Fodquake, and anyone I know who has tried it swears by it.

who ever plays load of crap FTE anyway?

fix FTE first.

lol @ 'cry me a river' :-)

Are we talking about downloading maps and demos?

U knew any engine which have zero bugs after release? "cl_nopext 1" will help, but here u start complain how this disturbing and "breaking" normal FTE servers.


I will, some day.


So chunked donwload must not be affected by ping, but with 12ms normal download method is faster.

Yes.

No, but I do know people who test their code prior to committing it to source control and who test new features prior to a new release, especially when implementing said feature could break a lot of stuff when done wrong. But who am I kidding? MVDSV has a long history of doing exactly the opposite. Committing random code to source control, breaking stuff, not testing then making a release.


Sounds reassuring.


At 12ms, the normal download method might be marginally faster, but at anything else, chunked downloads range from being faster to being a fuckload faster.

But OK, if you don't want it, I'm not forcing you to implement it. But at least don't release new versions of MVDSV which break all clients which do support it.

Looks like bigfoot is right this time... (if it is like he says)

i guess they test primarily with mvdsv+ezquake, mvdsv+zquake or mvdsv+qwcl, hence covering 99% of users

not many people test their javascript on Safari, this doesn't necessarily mean they are unprofessional, but simply that they are spending the time allocated to testing in the manner in which it is most likely to be used

instead of throwing your toys out the pram on forums...go post bug reports on SF indicating which build it last worked and the build when it stopped working, also include the build(s) of FTE you are using ;p

"I know this song." (c) Zorg
And I know such ppl too, Microfost for example, they test they code, they have RC1 RC2... etc.
Also it very ez blame when u have zero own projects, at least zero projects which have more than 2 ppl which used it. Its hard to break anything when noone used what u do.

Yes, this chunked download was't tested comprehensively, but it was tested. And problems was expected, but I am not cared(YES) about it since we have cl_nopext 1 and here not much client which support this feature. Really here was't need to include this feature in official release, that my fault. Still that not cricial, noone will die soon due to that, I promise.


Too late, we will have a baby, read above.

You know, some of the server admins blindly upgrading their MVDSVs

1) do not read changes from cvs
2) do not require the fixes that have been made
3) are not MVDSV/feature testers (eztv/fixes) and
4) do not care that the majority of the players can play at working servers.

I have blindly upgraded (almost) to every recent CVS mvdsv and I will continue to do that. I don't care if some client's download gets broken (cl_nopext 1??) by the upgrade if 1) the server crashes on mm1 2) server crashes with eztv 3) server crashes on player disconnecting on map_end 4) etc...

Majority of users still use MQW/Fuh/EZQ.

"There is a bug in last stable MVDSV version causing X not working when using Y."
That was a way to say it too, but then your topic doesn't get so many reply posts

What is fodquake? Not even google knows it!

I'd say you can easily expect them to only test MVDSV and Ezquake.

BTW, less than 1% of QW players use Fuhquake? Since when?


No, but if MVDSV already had Javascript support which worked fine with Safari, and MVDSV then added SafariScript support which completely broke MVDSV with Safari, then there's something wrong. Your analogy is completely flawed.


I don't use MVDSV. It works on servers running 0.25, it doesn't work on servers running 0.26. 0.25 doesn't have any support, broken or not, for chunked downloads.

I actually only tested with FTE to make sure it wasn't a problem with my implementation, but my FTE is the current from SVN.

If you're referring to me, I can assure you that I'm working on several things which have more than 2 persons working... And where I do try to make an effort to not have every single release be unusable due to added bugs not found due to missing testing.


Noone will die, but some people can't download demos from servers run by admins who decided to upgrade to this release. And then some servers might stick with this version for a while, etc, etc.

Ask yourself how come these problems are there in the first place. Maybe if you think hard about it, you'd have a revelation.

BTW, do you, as a server admin, feel comfortable running software which contains this amount of serious problems?

I've tried that approach more than one time before, and all I got was the usual stupid/useless/arrogant responses, so please excuse me for being cynical about MVDSV these days.

It's mentioned two times on the first result page, not bad for something which hasn't had any publicity and isn't even released yet.

In short it's a Quakeworld client which focuses on sucking less, security and working better with non-Windows operating systems as well as bringing general improvements, including support for some of the more useful FTE extensions.

When you make software, it's bound to have bugs/errors/whatever. If noone tests these builds, would the bugs be found at all? I'm willing to take shit from users if my server bugs somehow due to it being (unstable) CVS build. If I do this, I can report the problems for devs who will fix the damn thing eventually.


All of the problems I have encountered or people have told me have been fixed because I've been in touch with the developers. Only chunked download seems to be broken at this time (plus fast join/observe vip bug).

So yes, I feel comfortable running CVS builds because I know I can get bugs present in fixed by just talking to devs and not complaining about it (or bashing devs) on public bulletin board.

The progress so far has been outstanding, as most of the security holes has been fixed in the past year. Do you really want to blame people who do this for free and are willing to spent their FREE TIME for other people who mostly are not grateful what possiblities they have been given instead of being constructive and try to get problems solved?

I'm sorry to say it, but I was expecting to receive a URL as reply to my question, not propaganda!

to make a useless response
<3 bigfoot

go bigfoot

Biggie <3