These tests were done on the following system:

OS: Windows 7 RC (build 7100), 64-bit US English version
CPU: AMD Phenom II 955 quad-core @ 3.2 GHz
GPU: XFX Radeon HD 4770 512 MB GDDR5
RAM: 4 GB DDR3 dual channel unganged mode
HD: Samsung HD103UJ 1TB SATA2 (Windows write caching enabled)
Sound: Creative SB Audigy 2 with 5.1 surround sound
Xfire: tested both 1.119 and 1.120 beta (build 41000)
Game: Audiosurf
Game resolution: 1680x1050 (fullscreen)

Xfire video capture settings:
Frames per second: 30
Full-size
Capture cursor: enabled
Record sound: enabled
Audio Device: Vista Audio
Audio Mixer: Vista Audio Mixer
Enable Flashback Capture: disabled

Resulting video file:
Video length: 4:47 (30.000 fps)
Audio length: 4:32 (1 channel, mono)


To start, the resulting video file tells an obvious problem, the audio track in the resulting AVI file was 15 seconds shorter than the video track. Neither of those are correct, audio is too short and video is too long, however I don't remember the exact time I recorded the video.

The video recording was started on the track fly-by when the music is not yet playing. It took some seconds after the video capture starting to get into actual gameplay and when the music starts playing in the game. In the resulting video file the video data correctly started from the moment I hit the recording keys. The audio track, however, started from the point the game actually started playing music.

This itself caused the audio track to be several seconds ahead of the video track. Xfire only started recording audio when the game started playing it, but in the video there is no padding silence in the beginning, instead the audio starts immediately. This is a bug related to how Vista Audio works and how Xfire records sound data from it. If it has no audio available, it should record silence instead of not recording anything.

To work around that problem I cut the beginning of the video away so that the video starts from the point the audio started playing in the...

So I make my websites in XHTML. The correct type to serve them would be application/xhtml+xml. Works fine in most web browsers. Our old favorite, Internet Explorer, however is the odd exception to the rule. It has no idea whatsoever what type that is and instead of displaying the web page it asks if you want to save the file of "Unknown type".

Fix would be serving the page as text/html to fool IE thinking it is regular HTML document... let's get back to this in a bit.

A good thing that web browsers can supply Accept header in HTTP requests to tell which document types they accept and prefer. That way I can check if the browser prefers text/html more than application/xhtml+xml and then automatically serve the documents as html instead of xhtml. Still when I tested it, IE failed to recognize the file.

The reason? Let's take a look at the Accept field different web browsers send:
Firefox: "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8"
Opera: "text/html, application/xml;q=0.9, application/xhtml+xml, image/png, image/jpeg, image/gif, image/x-xbitmap, */*;q=0.1"
Chrome: "application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5"
Internet Explorer: "*/*"

So Firefox says it wants text/html or application/xhtml+xml. If neither is applicable, application/xml would be next. If the content cannot be sent as that either, then anything goes.

Opera prefers html and xhtml along with many image formats, then xml and last anything else.

Google Chrome actually specifies that application/xhtml+xml, application/xml and image/png would be the preferred forms, only then text/html, then text/plain and only then the rest of document types.

What about IE? It simply says "give me what you want, I accept it all with equal preference". Yet obviously it doesn't support all document types since application/xhtml+xml gets "Unknown type" download dialog...

So instead of doing it right and...

Some demotivational poster said "When you need it done right, do it yourself". Of course in that context it was referring to masturbation, but sometimes you just need to apply the same logic elsewhere. After waiting for several years for Xfire website to have some kind of feeds for blog comments, video comments and community news comments, and now even the main forums no longer can alert you on new replies, it seems like I need to do something about it myself. (I'm not blaming the Xfire web devs for not having those, though. Would be quite a lot of work for such a little target audience)

There are two ways I thought of how to get notification of new replies:

1. Create a news feed of it that can be subscribed to like to any Atom feed
   
2. Make an Xfire bot to message you of the new replies

This is where you come into play. Which way would you prefer? To get the news in your feed reader, or to get the messages by Xfire?
Also, which of the subjects mentioned above would be the most important one to get information from? (blogs, videos, news, forums?)

Of course I do this all for myself, but since it can be turned out to public for others to have benefit too, why not do it? Vote your preference by commenting on this blog post. I'll try to check the comments every once in a while...

I guess I have the pleasure to announce that today Xfire suddenly had an increase in the number of official gaming communities. One of those is very special to me, the Official Left 4 Dead Xfire Community.

If you wonder what makes it so special other than that I have played that game a lot lately, then join the community and find out!



Thanks!

So making my bug tracker I encountered a problem with MySQL. I have a view that holds information of bugs such as "last time edited".

For Atom feeds I needed to get the name of the person reporting the bug. Well, I thought of just adding this information to this view as it would make sense there. Trying my theory with an SQL query showed promising:

Quote

SELECT
bugid,
COUNT(*) AS historycount,
MAX(time) AS historytime,
(SELECT who FROM history WHERE bugid=bugid ORDER BY time ASC LIMIT 1) AS author,
(SELECT who FROM history WHERE bugid=bugid ORDER BY time DESC LIMIT 1) AS editor
FROM history
GROUP BY bugid

It worked nicely, for each bug it listed the number of edits done to it, when the last edit was made, who was the author (first history event for the bug) and who was the person doing the latest edit.

Okay, so those subqueries there worked great to achieve what I wanted. Now to apply them to the view where I could query the data from:

Quote

CREATE VIEW `bug_history` (
bugid,
historycount,
historytime,
author,
editor
) AS SELECT
bugid,
COUNT(*),
MAX(time),
(SELECT who FROM history WHERE bugid=bugid ORDER BY time ASC LIMIT 1),
(SELECT who FROM history WHERE bugid=bugid ORDER BY time DESC LIMIT 1)
FROM history
GROUP BY bugid

When selecting * from that view, everything looks the same as in the query itself... except for the "editor" column. Both editor and author columns had the same values, the name of the author. It seemed weird why this view would give different result with identical query.

After I read a few MySQL documents and tried changing ALGORITHM and playing around different variations, I came to conclusion that it could be a bug in MySQL. So I searched a bit through the MySQL bug database. After a while I found the answer: This is not a bug, it's working as intended.

While it doesn't really make sense for the VIEW to give different result than SELECT query...

If you want to add me to your friends list, first make sure I am not playing a game. Xfire still has not fixed the current behavior that friend invites in game override all other functionality (you cannot do anything with XIG before responding to the invite) and if you go out of the game, the invite is automatically denied.

And as I always look up information of the person trying to add me, it means your invite will be automatically rejected by Xfire if you send it while I am in game.

I'm also waiting for the day when one can use Xfire when a friend invite is pending. As it is now, a friend invite blocks you from using Xfire at all. Only the existing chat windows are accessible but the main interface is locked.

So if your friend invite got rejected, don't feel bad about it. I just want to use Xfire and your invite prevented me from doing that.

(PS. If you actually want a chance of your invite being accepted, remember to state some kind of reason for your friendship request. "Add me to your friends list" won't cut it.)

I wonder when Xfire can manage to change the forum software. The current phpBB 2 hack-up is failing in so many ways. And I know I am not the only one thinking so.

It has happened to me more than a few times that after I post a message I notice a little typo so I go edit the post and submit the fixed version. The forum software would complain "You are not allowed to make another post so soon." But I was not making another post, I was fixing an existing one! So in other words the forum software does not want people fixing their mistakes quickly.

Then another thing: when you post a new reply and double-click the Submit button. It allows that and does actually post two identical replies with the same submit time! So you are not allowed to edit your post quickly, but you are able to post multiple replies in rapid succession? Doesn't this seem a bit backwards?

I guess you could call this to be one of the "GJ only" bugs. http://godjonez.ath.cx/~godjonez/jing/2009-10-06_1019.png

So I was "online" in Xfire to my account but the only friends online were for my AIM and MSN accounts, none of the Xfire friends. Amazing, isn't it?

Well, if I can get weird bugs happen, so can other people too. Or what do you think of a case that sending a text message in Xfire chat breaks up computer monitor by causing a physical crack on it? http://www.xfire.com/xf/modules.php?name=Forums&file=viewtopic&t=254331