Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
Dacicus wrote:
I was able to get Flash Player running in DSL, but it was the Firefox plug-in of version 7.
I searched the Red Hat install for those dependencies linked above and found that the following are missing for FP9:
libfreetype.so.6
libfontconfig.so.1
libgtk-x11-2.0.so.0
libgobject-2.0.so.0
libglib-2.0.so.0
glibc>= 2.3 (has 2.1.3)
support for V4L1 (not sure how to check this)
For FP10, the following are also missing:
glibc>= 2.4
libnss3
cURL
support for V4L2 (again, not sure how to check this)
Checking V4L[1|2], check if your kernel was compiled with Video4Linux support.
To know which version of Video4Linux that's available, a quick google gave me this "Version 2.6.27 of the Linux kernel dropped support for the legacy Video4Linux (1) API." (from https://wiki.archlinux.org/index.php/webcam_setup ).
The configurations needed for each kernel can be found in this list: https://cateee.net/lkddb/web-lkddb/index_V.html
It's easiest to check if this is enabled by using "gunzip /proc/config.gz | grep [config option goes here]", this assumes /proc/config.gz is enabled.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
A little more information on why it's not working will help us figure out why it's not working for you.
Are you getting any error messages? If so, which?
How are you loading the ROM?
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
Nach wrote:
Also, maybe we should consider VirtualBox-RR.
VirtualBox-RR? I am not touching anything Oracle has had their hands in. Do you recall my whine about their MySQL drivers? Until someone proves otherwise this is how I see the Oracle development process: https://xkcd.com/1926/
(I've also used VirtualBox for a lot of things, the number of times it cost me my foot rather than won me anything is pretty high)
Patashu wrote:
Sounds like improving keylie’s Linux TAS tools ( https://github.com/clementgallet/libTAS ) then in the future TASing windows games by emulating them through TAS’d Linux is the way to go, lol
There are way better methods on Linux actually, the kernel is quite a bit more powerful in regards to these things than Windows is.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
This was definitely not the way I saw the project going when I decided I wanted to jump in and improve on nitsujas work. Most who are reading this are probably now asking "How did we get here?". I will try to cover that.
Final chapter serves kinda as a tl;dr, for you lazy people.
Basically, there was some information that I had overlooked. I received a tip on IRC regarding a book called Windows Internals, I believe it was Henke37 who suggested it, and this book is full of nightmares.
Besides showing me Hourglass isn't the way to go for re-recording Windows programs, it also turned me off on Windows entirely.
How bad could it really be?
There are some points discussed in this book that just sank my heart. Could windows be lying? No way! Yes way!
They have multiple shim-engines in their kernel that patches old programs at runtime to make them run better. They are also lying to debuggers about when a program has started, when DLLs have been loaded, etc.
This becomes problematic for a debugger such as Hourglass, since it needs to know these things when they happen, not when the kernel thinks it needs to know these things.
The straw that broke the camels back
To mitigate the troubles we'd need a kernel driver. This driver would poke at structures that the book Windows Internals doesn't describe (though, they are mentioned), and the Microsoft kernel debugger won't print full details of. This feels like a horrible idea, and also needs an additional computer as Windows can only be debugged over Ethernet for these things. I am not interested in buying another computer just for this.
This just to get an idea of when a program really begins to start, memory allocations, thread execution orders etc.
There would be drivers Microsoft aren't going to approve of. Which becomes annoying really fast since Windows 10 only accepts signed drivers, outside of "hidden" debug settings.
But, WinDbg has re-recording?
Yes, it turns out that WinDbg has added re-recording, and great for them. They avoid several issues Hourglass is facing by recording program state rather than input.
I did not research this a lot, so I am not sure how they solved some of the problems that we'd have in common with WinDbg, such as memory allocation handling.
"We're doomed! Doooooooooooooooomed!"
Not quite, there is at least one kernel that provides a lot of this control, and contains a fairly well documented API, the Linux kernel. There's also a program that enables the launching of Windows applications on Linux, Wine.
There's a project that I got to see the insides of that looks really promising in terms of making this happen.
Finally, I want to say thank you to everyone who contributed. I hope some of you will come back when the next project is up and running.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
Well done, glad to see that you finished it.
Entertainment-wise I'm with Spike, the GBA games are not that fun to watch (but fun to play), so I will be voting meh.
I would really like to see more submission text, explaining the tricks, movement tech, etc.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
I'm going to re-iterate: Repetition is not the King of Iron Fist. Repetition is not the King of Iron Fist. Repetition is not the King of Iron Fist.
There's also no point in going in for the overkill unless you want to show something specific off that requires it, otherwise save it for another fight.
Looking into this more, no character has enough interesting moves to warrant a full story mode play-around, except for Combot (which can be most characters), although due to limitations of Combot in Tekken 4, you may need to do 1 round battles and run the story mode twice, as he switches simulated character per match instead of round.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
And this is why google translate is insufficient as communication. Much like we occasionally help Japanese TASers by allowing them to post in Japanese and later translate, if you would have asked for help, another French member would have helped you understand Spike's post.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
I do not agree with removing Warcraft II from Stars. It's our only starred RTS TAS, and in my opinion is a perfect example of RTS TASing, with AI mockery and playaround.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
11:53 <feos> ThunderAxe31: hi do you have a linux installation capable of running libTAS?
11:54 <ThunderAxe31> I only have an old Linux Mint version for USB live booting
11:54 <ThunderAxe31> why you ask?
11:54 -!- feos [~feos@95-25-108-126.broadband.corbina.ru] has quit [Read error: Connection reset by peer]
11:55 -!- feos [~feos@95-25-108-126.broadband.corbina.ru] has joined #tasvideos
11:55 <ThunderAxe31> feos: <ThunderAxe31>I only have an old Linux Mint version for USB live booting
11:55 <ThunderAxe31> <ThunderAxe31>why you ask?
11:55 -!- feos [~feos@95-25-108-126.broadband.corbina.ru] has quit [Read error: Connection reset by peer]
11:55 <ThunderAxe31> ERROR 666: ANSWER SUCKS
11:56 -!- feos [~feos@95-25-108-126.broadband.corbina.ru] has joined #tasvideos
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
EZGames69 wrote:
Nonster wrote:
Thank you for many replies.
By the way, I got new TAS record in 25:02.1.
Should I post again?
I think you can just upload your new movie to userfiles and a judge will replace it for you.
It's linked in the submission already and the judges are aware. The submitted movie will be updated as soon as a judge has time.
There is no need to post the movie again Nonster.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
ThunderAxe31 wrote:
Despite wasting a total of about 15 seconds between fight transitions, I still think this submission is worth of Vault, for two reasons:
First, optimizing menuing is a marginal factor that doesn't consist in actual TASing.
Second, the Millenniummon fight, which is the only speedrunnable part of the game, is clearly superhuman, and even if we add these 15 seconds to it, for a total of 44 seconds, it still beats all known RTA records, and in my opinion is also still impossible to match in real time attempts.
Technical requirements
- Must beat all known unassisted records and beat or match all known tool-assisted records.
- Must show a proficient understanding of TAS techniques and be perceived as well optimized. I.e. should not have obvious pauses, mistakes, or sloppy play.
- In rare cases, there will be a sloppier movie that is faster than a more optimized movie due to the use of a major skip discovery. In this case, the faster of the two movies is preferred.
- The guidelines for difficulty settings, and password use still apply.
- Must be distinguishable from the best real-time speedruns, and must not be seen as trivial.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
I am also in favor of creating this new award.
Technical TAS sounds like it would be easy to misunderstand and give to runs of high technical value, even if it does not come from research. Same with the suggested icons.
For example, the above suggests a TAS like the most recent Pokemon Yellow ACE would qualify, which is (as far as I gather from the posts here, and from discussions on IRC) not the goal of the award.
My proposals for name and icon:
Investigated TAS, magnifying glass over a game cart.
Planned TAS, an "X marks the spot" map with paths dotted on it.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
Some nice combos in there, but since you went for a play-around, I want more silly-ness like choreography with the music, use of moves that are "guaranteed" to get you owned if attempted when playing normally, etc. Also "impossible" combos, as many different as possible shall be shown off. Repetition is not the king of Iron Fist.
The Tekken 4 and 5 games have some interesting visual panics as well. I don't recall any specific ones in Tekken 4, but look for them, and trigger them. An example is Tekken 5 where you with Feng can start some of the special grabs when the opponent's character model is in an unexpected position (such as Law's Frogman), and the game compensates for it by just teleporting the opponent into a functional position.
Also, RIP sound emulation....
I did enjoy the grab-battle with Christie, and the occasional attack that's not meant to connect.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
http://problemkaputt.de/gbatek.htm#gbamemorymap
I forget which domain is which, but I would assume the On-chip work RAM being IWRAM, and the On-board work RAM being EWRAM. Test both for all values, not just one.
If no value ever changes, first make sure you're using the same region ROM.
Then, if you still have a problem, try to fetch it from Lua, there are lots of examples packaged with Bizhawk, you can start from one of them. You can also reference to this: http://tasvideos.org/Bizhawk/LuaFunctions.html
If Lua works and RAM Watch doesn't, report a bug (if the issue is still present in the latest release). Otherwise you definitely have the wrong addresses and you need to find everything again.
Joined: 3/2/2010
Posts: 2174
Location: A little to the left of nowhere (Sweden)
If you want the final encode to contain the subtitle commentary, provide them as a subtitle file (e.g. .sub, .srt or .ass).
Our encoders are most likely not going to spend the better part of several months to watch your youtube videos, type out your commentary by hand, and then re-work all the timing values. Which is an extremely unreasonable demand to make.
