Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
On Sat, Jul 31, 2010 at 6:48 AM, buginator wrote: ... > On another note, there is another issue, that has to do with TCP, and > how it blocks. > The fix for this isn't so easy, even setting all sockets to > non-blocking, and use polling to determine if done won't solve all the > blocking issues that can happen. > > I don't think newnet does anything different in this area either. > > We could spawn a thread for each player and have a main network > handler event loop that queues everything up to prevent blocks, but > this isn't easy to code, and would pretty much force us to enter beta > mode again (2.4 beta 1 or ??). > Other ideas is, use raknet, or switch to UDP, but those also force us > to enter beta mode as well. > Other options ? Actually, newnet does do something different in this area. Newnet writes from a separate thread, using polling to determine when writing is possible. This was actually needed to be able to send desynch logs without freezing. Switching to UDP might make sense as a way to reduce bandwidth usage (think UDP might have less overhead than TCP, not sure), but not as a way to prevent blocking. -Cyp ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
On 7/27/10, Per Inge Mathisen <> wrote: > Let's keep our cool for a bit and see if more bloopers surface before > we rush to re-release anything. > > Then we can push out 2.3.3 next weekend, hopefully this time with some > actual testing done. Although I cannot promise anything in that > department -- Starcraft 2 just came out. > Just a note, the blockers that are left are: http://developer.wz2100.net/ticket/2029 (trivial fix, didn't check for player 9 (features)) and http://developer.wz2100.net/ticket/2037 (which is a big issue, since people can't continue the SP game.) For the fix on this one, we can either revert all fixes, and the original patch, http://developer.wz2100.net/ticket/1971 or we won't be able to do a release tomorrow. Zarel is working on this though, and perhaps he can come up with a fix. Anyone else know of anything else that is still a blocker ? On another note, there is another issue, that has to do with TCP, and how it blocks. The fix for this isn't so easy, even setting all sockets to non-blocking, and use polling to determine if done won't solve all the blocking issues that can happen. I don't think newnet does anything different in this area either. We could spawn a thread for each player and have a main network handler event loop that queues everything up to prevent blocks, but this isn't easy to code, and would pretty much force us to enter beta mode again (2.4 beta 1 or ??). Other ideas is, use raknet, or switch to UDP, but those also force us to enter beta mode as well. Other options ? ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
On Tue, Jul 27, 2010 at 6:31 PM, buginator wrote: > On 7/27/10, Per Inge Mathisen <> wrote: >> Then we can push out 2.3.3 next weekend, hopefully this time with some >> actual testing done. Although I cannot promise anything in that >> department -- Starcraft 2 just came out. > > Per, I expect a review :) And I expect you to change your mind about direct debit after trying StarCraft II. ;) > On 7/27/10, Guangcong Luo <> wrote: >> On Tue, Jul 27, 2010 at 3:23 PM, Stephen Swaney <> wrote: >> > Just one more example of why programmers are the worst people to test >> their >> > own code. >> > > Erm, well, that's true, but what's your point? > > The point is, that we don't have actual testing being done, except by > the person that wrote the patch. Well, yes, but we already know that. Pointing it out again without suggesting a solution seems unnecessarily accusative. > We don't have any reliable beta testers, or a way to test new patches > that will cover most all cases, as can be seen by the current issues > with 2.3.2, and the other 'hotfixes' we had in the past. > > Having it in trunk first, slightly helps, but since that version has > become unstable for testing, we are more or less in a bind. I'd argue having it in all branches first will increase the likelihood that it is tested and that bugs are found. Of course, not everyone seems to agree, so so far I've been doing exactly that - committing larger patches to trunk first. > This is also why I have been wanting to have a prominent location next > to the green (:P) download button, for test builds--assuming we can > work something out with Fastdeath, and have nightly builds available > again. I'd suggest that any build linked from the front page be at least alpha quality - no nightlies. The main reason for this is that we want people to test multiplayer, and that gets difficult if the build changes every day. I'd also prefer that whenever a build is released, it's intentional, so that when there's a known bug that makes the game unplayable, we can fix it without a bunch of people telling us that our testing build doesn't work. Remember - the front page is a _very_ prominent location. > Even if we do have nightly builds, the sad fact is, that most people > don't want to test builds, they rather just use a 'stable' build, so > it might just end up a wash either way. You've been saying that a long time, but I don't believe it is true. During the 2.3 beta cycle, everyone in -games was on 2.3, and the 2.3 lobby was significantly more active than the 2.2 lobby. Many people, if given a choice between new features and stability, would choose new features. For instance, I played trunk whenever possible (until power queue was committed; I've been playing it less since then). > What is the answer to this ? Prevent any new features from going into > 2.x, and just make it a bug fix only branch ? > I don't think that will work for us, but I am out of good ideas. I think releasing testing builds is a fine idea. Nightlies, too, as long as they aren't broadcast on the front page. -Zarel ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
On 7/27/10, Per Inge Mathisen <> wrote: > Let's keep our cool for a bit and see if more bloopers surface before > we rush to re-release anything. > > Then we can push out 2.3.3 next weekend, hopefully this time with some > actual testing done. Although I cannot promise anything in that > department -- Starcraft 2 just came out. Per, I expect a review :) On 7/27/10, Guangcong Luo <> wrote: > On Tue, Jul 27, 2010 at 3:23 PM, Stephen Swaney <> wrote: > > Just one more example of why programmers are the worst people to test their > > own code. > > Erm, well, that's true, but what's your point? The point is, that we don't have actual testing being done, except by the person that wrote the patch. We don't have any reliable beta testers, or a way to test new patches that will cover most all cases, as can be seen by the current issues with 2.3.2, and the other 'hotfixes' we had in the past. Having it in trunk first, slightly helps, but since that version has become unstable for testing, we are more or less in a bind. This is also why I have been wanting to have a prominent location next to the green (:P) download button, for test builds--assuming we can work something out with Fastdeath, and have nightly builds available again. Even if we do have nightly builds, the sad fact is, that most people don't want to test builds, they rather just use a 'stable' build, so it might just end up a wash either way. What is the answer to this ? Prevent any new features from going into 2.x, and just make it a bug fix only branch ? I don't think that will work for us, but I am out of good ideas. ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
On Tue, Jul 27, 2010 at 3:23 PM, Stephen Swaney wrote: > Just one more example of why programmers are the worst people to test their > own code. Erm, well, that's true, but what's your point? -Zarel ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
At least they can write codeTesting should be done on SVN Branch 2.3? On Tue, Jul 27, 2010 at 15:23, Stephen Swaney wrote: > On Tue, Jul 27, 2010 at 02:51:44PM -0500, Guangcong Luo wrote: > > > > To be fair, it only breaks loading maps in T1 for some maps, and Rush, > > the map I usually test with, worked just fine. Anyway, I've fixed that > > too, now. > > Just one more example of why programmers are the worst people to test their > own code. > > And why "I tested it!" is a terrible reason to allow anyone to add > brand new code to a release at the last minute. > > -- > Stephen Swaney > sswa...@centurytel.net > > > ___ > Warzone-dev mailing list > Warzone-dev@gna.org > https://mail.gna.org/listinfo/warzone-dev > ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
On Tue, Jul 27, 2010 at 02:51:44PM -0500, Guangcong Luo wrote: > > To be fair, it only breaks loading maps in T1 for some maps, and Rush, > the map I usually test with, worked just fine. Anyway, I've fixed that > too, now. Just one more example of why programmers are the worst people to test their own code. And why "I tested it!" is a terrible reason to allow anyone to add brand new code to a release at the last minute. -- Stephen Swaney sswa...@centurytel.net ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
On Tue, Jul 27, 2010 at 7:36 AM, Christian Ohm wrote: > On Tuesday, 27 July 2010 at 3:20, Guangcong Luo wrote: >> 2. Starting a game in T2 or T3 mode doesn't work - the game just >> starts in T1 mode. > > r11305 breaks loading maps in T1. To be fair, it only breaks loading maps in T1 for some maps, and Rush, the map I usually test with, worked just fine. Anyway, I've fixed that too, now. -Zarel ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
On Tue, Jul 27, 2010 at 3:22 PM, Per Inge Mathisen wrote: > next weekend I mean this weekend. - Per ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
Let's keep our cool for a bit and see if more bloopers surface before we rush to re-release anything. Then we can push out 2.3.3 next weekend, hopefully this time with some actual testing done. Although I cannot promise anything in that department -- Starcraft 2 just came out. - Per ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
Re: [warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
On Tuesday, 27 July 2010 at 3:20, Guangcong Luo wrote: > 2. Starting a game in T2 or T3 mode doesn't work - the game just > starts in T1 mode. r11305 breaks loading maps in T1. ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
[warzone2100-dev] Release 2.3.2a or 2.3.2.1 or 2.3.3?
Hey, guys. Two fairly bad bugs kind of went into 2.3.2 and no one found them. :( They are: 1. When a fully upgraded object takes burn damage, it's instantly killed (underflow in damage calculation). 2. Starting a game in T2 or T3 mode doesn't work - the game just starts in T1 mode. I've fixed both of these in 2.3-branch. Should we release a fixed build immediately? -Zarel ___ Warzone-dev mailing list Warzone-dev@gna.org https://mail.gna.org/listinfo/warzone-dev
