0.2.8_beta4: release process and bugs
0.2.8_beta4: release process and bugs
Tagged the sources. The name of the tag is v0_2_8_beta4. I'll do some more tests tomorrow. Consider the tag movable for now, until a source tarball bubbles up.
I had this happen a few times--your camera stops following you. It happens after many 180s (seemed to happen more frequently at the rim wall). It happens at the end of the recording (~380).
Edit: Comma was missing from the default word delimiters, I added it in on b0_2_8.
Edit2: Maybe only '_' and space should be default. No idea. Anyone have a list of default delimiters in text editors?
Edit: Comma was missing from the default word delimiters, I added it in on b0_2_8.
Edit2: Maybe only '_' and space should be default. No idea. Anyone have a list of default delimiters in text editors?
You do not have the required permissions to view the files attached to this post.
I'll update the tag to the new delimiters. No idea about defaults here, either; I'd just use whitespace, but I see how _ would be useful.
About the recording: It's working fine here, no visible glitch. Your cycle stays in a fixed position on the screen, and for the framerate to turn rate ratio, it reasonaby stays behind the cycle. Can you send screenshots, too?
About the recording: It's working fine here, no visible glitch. Your cycle stays in a fixed position on the screen, and for the framerate to turn rate ratio, it reasonaby stays behind the cycle. Can you send screenshots, too?
Problems with the Windows build. The VC6 executable has the random-characters-flickering-during-chat problem again, so probably the string is not properly NULL-terminated. And I saw Fonkay's random camera focus switching bug once (of course, without recording it).
I'm already tired, so no release today. Sorry.
I'm already tired, so no release today. Sorry.
Jonathan mode:
Guess the result of In VisualC 6
Guess the result of
Code: Select all
unsigned int x = 0xFFFFFFFF;
float f = float(x)/float(x);
std::cout << f << "\n";
No, but pretty close.
The camera switching bug is a multithreading bug and completely mine. I temporarily set the living status of cycles to zero while I move them. The sound system gets the camera focus, and the called routine will switch the focus once it notices the focus object is dead. Duh. This could fill several chapters in a "Coding practices to avoid" book. I'll have to fix this tomorrow, no chance I can wrap my mind around it now.
The camera switching bug is a multithreading bug and completely mine. I temporarily set the living status of cycles to zero while I move them. The sound system gets the camera focus, and the called routine will switch the focus once it notices the focus object is dead. Duh. This could fill several chapters in a "Coding practices to avoid" book. I'll have to fix this tomorrow, no chance I can wrap my mind around it now.
- Lucifer
- Project Developer
- Posts: 8642
- Joined: Sun Aug 15, 2004 3:32 pm
- Location: Republic of Texas
- Contact:
Here's a recording of the disappearing console messages bug I've been having. At about T860 in the recording, caretboy ( _^_ ) asked "is it defence or defense?". I answered "defense", but it didn't appear. He said "thank you" like he read it. I couldn't get him to tell me whether or not he saw it, but I did type it.
The other two bugs I was hoping to catch red-handed didn't appear: the disappearing zone (happened earlier) and the weird crash.
If you watch the first part of the recording, you'll see people resonding to some of the new default instant chats. Since people were offended, and they're the "I'm an ass" chats, I think you can say "mission accomplished". Heh.
The other two bugs I was hoping to catch red-handed didn't appear: the disappearing zone (happened earlier) and the weird crash.
If you watch the first part of the recording, you'll see people resonding to some of the new default instant chats. Since people were offended, and they're the "I'm an ass" chats, I think you can say "mission accomplished". Heh.
You do not have the required permissions to view the files attached to this post.
- Lucifer
- Project Developer
- Posts: 8642
- Joined: Sun Aug 15, 2004 3:32 pm
- Location: Republic of Texas
- Contact:
Potential bug, somewhere from 8:45-8:55 (pm, CST). I forgot the time already. I ws in the server, and a "Lucifer" joined the server, and the name looked colored just like mine. I don't know if it was an imposter or not, a simple grep of the log should tell you.
I said "BUG"
Then "BUG - " something about explaining the other Lucifer. So a grep with 5 lines of context on the log should turn it up, he left right before I made the BUG call.
Um, this is on Fortress.
Edit: added pm and timezone
I said "BUG"
Then "BUG - " something about explaining the other Lucifer. So a grep with 5 lines of context on the log should turn it up, he left right before I made the BUG call.
Um, this is on Fortress.
Edit: added pm and timezone
Last edited by Lucifer on Sat Nov 19, 2005 7:11 am, edited 1 time in total.
- Jonathan
- A Brave Victim
- Posts: 3391
- Joined: Thu Feb 03, 2005 12:50 am
- Location: Not really lurking anymore
OK, I'll try the other side.z-man wrote:Jonathan mode:
Sounds like something is really screwy. My guess is that the optimizer is brain-damaged again, but I don't know what it did. Perhaps it has problems with unsigned integers, casts, or duplication? My guess for the result: NaN, infinity, or some odd number (not necessarily as in not even).z-man wrote:Guess the result ofIn VisualC 6Code: Select all
unsigned int x = 0xFFFFFFFF; float f = float(x)/float(x); std::cout << f << "\n";
ˌɑrməˈɡɛˌtrɑn
Lucifer: This is going to get interesting The recording plays back smoothly. When it comes to the point where your chat message got lost for you, the message is displayed flawlessly.
What is your exact compiler version and the exact CXXFLAGS you're using? I'm trying standard debug (DEBUGLEVEL=3), standard optimize (no parameters) and lean debug (CXXFLAGS="-O0 -g") settings, and I'm currently letting it run without fast forwarding.
I'm not stopping beta4 for this. It's not frequent and obnoxious enough, and beta4 will help find it.
nemo: looks like texture lookup inaccuracies, like the ones you get when infinity is turned on. Is it? ATIs are not very good at real projective geometry Also, disabling the high rim may help.
Jonathan: your guessed result is pretty inaccurate You're not even excluding 1.
It's not an optimization failure, I noticed it first in debug mode.
Exchanging one F for a 7 and you get 1 as a result.
This should leave you with only two possible guesses
Also, splitting the operation into several lines produces 1, but that does not help your guessing. It does, however, show just how screwed up it is.
What is your exact compiler version and the exact CXXFLAGS you're using? I'm trying standard debug (DEBUGLEVEL=3), standard optimize (no parameters) and lean debug (CXXFLAGS="-O0 -g") settings, and I'm currently letting it run without fast forwarding.
I'm not stopping beta4 for this. It's not frequent and obnoxious enough, and beta4 will help find it.
nemo: looks like texture lookup inaccuracies, like the ones you get when infinity is turned on. Is it? ATIs are not very good at real projective geometry Also, disabling the high rim may help.
Jonathan: your guessed result is pretty inaccurate You're not even excluding 1.
It's not an optimization failure, I noticed it first in debug mode.
Exchanging one F for a 7 and you get 1 as a result.
This should leave you with only two possible guesses
Also, splitting the operation into several lines produces 1, but that does not help your guessing. It does, however, show just how screwed up it is.