Hard Light Productions Forums
Modding, Mission Design, and Coding => FS2 Open Coding - The Source Code Project (SCP) => Topic started by: Scuddie on April 26, 2006, 03:33:29 pm
-
OK, so we have alot of community based projects for SCP, and we have alot of files and versions associated with these projects. Lots of dependancies, and lots of documentation. There is so much going on, that the possibilities are endless. Yes, I'm saying it, FSO is nothing but a jumbled mess. It is in a complete state of disarray... Well, maybe not that far, but suffice to say it could use some housekeeping.
I cant count how many times I needed to play certain campaigns with different versions of executables, different MediaVPs, and different options. The whole thing is so messy, the current headlined release package took about 45 minutes to clean up just so I could use FSPort 3.0.3. If this had been a person unskilled at FS's native file formats, it would take hours of trial and error, and eventually giving up. Matter of fact, I actually had to help a friend of mine set it up, and it took about two hours for everything to work the way it should. And this is from a 'fresh package'. So now whenever I tell my unexperienced friends about FS2Open, I practically have to make a package myself and tell them not to update any of it unless I am there personally to supervise.
Yes the project is big and featurefull, but when was the last time a code freeze was instated? It seems like everybody's got the feature creep, and putting just as much effort implementing new things as they are fixing bugs. Beccause of this situation, official builds are few and far between. Between 3.6.6 and 3.6.7 there were dozens of CVS builds and custom builds. And as far as I can see, 3.6.8 is far, far away. I don't think this is supposed to happen, atleast dont see this as being practical.
Now don't get me wrong, I love all the work you guys are doing for the good of FSO, but right now I think all the skills, talent, and effort are misplaced. When I show someone the greatness of FSO, I don't want them to see the hundreds of different versions of the EXEs, nor the multitude of different versions of MediaVPs, nor the Wiki for help (which is in no way a Readme file), nor the workarounds and hacks for the many campaigns that dont work with different versions of FSO. I just want the gamer to be able to download a single official package, and have everything work from there.
With all the attention and publicity we've been getting lately, it would make sense to tidy everything up. Want to create a new mission flag? Great, just do it after the current mission parsing code works with all campaigns. Env mapping for OpenGL? That would be awesome... That is, after Env mapping for D3D is fixed. Really, things need to be fixed. Not only should things be backwards compatible, but also future compatible. That means implementing something, and then leaving it alone once it works. Once everything works, THEN you should try for something new.
Sorry for being so blunt, but I don't believe it makes any sense for everybody to do their own thing on a project this active.
Keep up the good work. And forgive the lengthy post.
[/rant off] :D
-
we're technically in a code freeze now. mantis would look alot better if PEOPLE WOULD KEEP US UPDATED ON THE STATUS OF THE BUG INSTEAD OF ABANDONING IT AND KEEP US GUESSING ON WHETER OR NOT THE BUG WAS FIXED. (edit) SPEAKING OF MANTIS, IF THE BUG ISN'T IN THERE THEN WE DON'T KNOW ABOUT IT. IF THERE'S A BUG THAT YOU DON'T POST ON MANTIS, AND THEN POST ON THE FORUMS THAT IT STILL HASN'T BEEN FIXED, I'LL BE EXTEMELY ANGRY -- PERHAPS ANGRY ENOUGH TO TYPE A POST IN ALL CAPS!!!!1111111
oops caps. wasn't directed toward anyone specific around here, but its just a general sentiment of mine when looking at mantis.
-
You may want to prioritize the bugs posted by people who are known to be around regularly. I have had three bugs on there for a few months now, one of which is fairly serious and that I would really like to see fixed (the support ship one).
Speaking of code freezes, I agree that it is probably a good idea at this point, but I don't want to say it publicly since there are three very minor features that I would like to see added in first. Oops, I said it. :p
-
You may want to prioritize the bugs posted by people who are known to be around regularly. I have had three bugs on there for a few months now, one of which is fairly serious and that I would really like to see fixed (the support ship one).
Well out of all the forum regulars not on the SCP team, you do have the most reported bugs. most of them have 1-2 reported usually.
-
http://lore.maxgaming.net/~scp/mantis/summary_page.php
reporter stats:
open/resolved/closed/total
kasperl 1 21 33 55
Goober5000 9 28 5 42
thesource2 11 13 14 38
WMCoolmon 4 27 7 38
CP5670 7 21 4 32
The Trivial Psychic 5 21 4 30
Lightspeed 3 14 6 23
karajorma 4 15 1 20
Hippo 3 12 4 19
diamondgeezer 2 9 8 19
-
Wow! I made the top 10!
-
Go, Scuddie, go!
Actually, when I followed the project a little more closely a while ago, I seriously got the impression that it would have imploded way back when if it hadn't been for taylor. All he seemed to do was patch over the holes other people created with supposedly great (but often enough apparently completely untested) new "features".
Not to step on anybody's toes, but that's obviously not really a sustainable development model. Not sure if the situation has changed since then, but looking at the forum postings, "feature creep" looks like a recurring theme...
(And it is insanely difficult to install some of the campaigns, though that's not necessarily a problem of the SCP (the insane number of intermediate/unofficial/whatnot releases is, though). "You need this from here and that from there, but make sure you don't get the package with the thingamajig, since that crashes on two builds my uncle found in a hollow tree just left of the third link in the last forum post, somewhere..." Argh!)
There, I said it aloud. *goes into hiding*
-
I cant count how many times I needed to play certain campaigns with different versions of executables, different MediaVPs, and different options. The whole thing is so messy, the current headlined release package took about 45 minutes to clean up just so I could use FSPort 3.0.3.
To be fair to ShivanSPS his first attempt at a big download had some errors he's hopefully avoided in his new repackage. We simply need to get it up and torrented.
It's also worth pointing out that he isn't a member of the SCP. Simply someone who saw that things could be improved who got off his arse and actually did it. That's what we need more of in the community.
If this had been a person unskilled at FS's native file formats, it would take hours of trial and error, and eventually giving up. Matter of fact, I actually had to help a friend of mine set it up, and it took about two hours for everything to work the way it should. And this is from a 'fresh package'. So now whenever I tell my unexperienced friends about FS2Open, I practically have to make a package myself and tell them not to update any of it unless I am there personally to supervise.
I agree that this needs to be changed but the problem is how do we change it? I've been pushing for a certain way of installing for over a year now. I've finally got some people to understand why running the media VPs as a mod is the best thing to do and why mod.ini files are so important but it's been one hell of an uphill slog. Every time something changes it takes ages to get everyone to accept the change and move on. I'm still seeing people advising the installation of the media VPs and even new campaigns in the main game folder! I still hear people suggesting the installation of things to Freespace2\Data :rolleyes:
It's pretty hard to clean up if the end users aren't willing to listen. For the most part I tend to advise now and let people choose to screw things on their own up if they don't want to listen to me.
Yes the project is big and featurefull, but when was the last time a code freeze was instated? It seems like everybody's got the feature creep, and putting just as much effort implementing new things as they are fixing bugs. Beccause of this situation, official builds are few and far between. Between 3.6.6 and 3.6.7 there were dozens of CVS builds and custom builds. And as far as I can see, 3.6.8 is far, far away. I don't think this is supposed to happen, atleast dont see this as being practical.
I tend to agree there. We do need to cut the waiting time between official builds down a bit. I think 4 releases a year might be closer to a good number than the current sort of schedule where we have maybe two official releases. It does pain me to have to recommend newbies get hold of a CVS build to solve problems because only SCP members and developers should be using those. We shouldn't be telling people who just want to play the game to use them.
That said maintaining an official build fork with bug fixes but no new features is a royal pain in the arse and I simply don't see any way that 3.6.8 isn't going to have bugs in it like the ones that make it necessary to upgrade to CVS builds. Unless someone steps up and is willing to take on the job of maintaining an official release build up to date I really don't see what we can do about it.
With all the attention and publicity we've been getting lately, it would make sense to tidy everything up. Want to create a new mission flag? Great, just do it after the current mission parsing code works with all campaigns.
Now here is where I start to disagree. The SCP's job is to make sure we are backwards compatible with the main FS2 Retail campaign. Nothing else. Everything apart from that is up to the community to keep current. Simple fact is that it is far too much effort to make sure everything works and often the problems are caused the developer of the campaign doing the wrong thing in the first place. Another large source of problems is the end user.
For instance look at Transcend. I must have answered the same question about 10 times on that thread because people would get crashes due to them not using an up to date CVS build even though it was expressly stated in the first post that you needed a build newer than a certain date. Other problems sometimes happen because someone has used bleeding edge features and then things have changed. If you must use the very latest CVS builds of FRED then you should realise that the feature set of those builds is subject to change. I'll definitely try to avoid changing anything that I've added especially if it is present in an official release but I reserve the right to change features if they really do need fixing.
Why is it up to the SCP to go through every single mission and campaign and make sure they all work? That's not our job nor should it be. Updating your campaign to keep it current is a pain but it's a minor one. Most of the flaws with mission parsing etc could be fixed if the FREDder or tabler involved went through their campaign and fixed them. Probably wouldn't take more than an hour or so in some cases. If you're not prepared to keep your campaign current you shouldn't be surprised if things go out of date and don't work with new builds.
Personally I'd suggest that when 3.6.8 is released a bunch of people go through every released campaign and make sure that they all work with it. Put up the fixed versions somewhere prominent and obfuscate the old links. I'd be more receptive to complaints about the inability to parse old missions if I knew that they had been fixed upto proper SCP standards in the first place at some point.
Env mapping for OpenGL? That would be awesome... That is, after Env mapping for D3D is fixed. Really, things need to be fixed.
Do you know how to program for D3D? Cause I sure as hell don't. Bob is probably the only person we have left who does. Quite frankly the team is very tempted to ditch D3D completely as it would make our lives easier. Until Bob comes back and starts working on it or someone joins the team who understands D3D I'd suggest that you consider the entire API depreciated. Hell the only reason I haven't suggested releasing Launcher 5.4 which can't select D3D as a means to hide it is because occassionally someone can't run OpenGL but manages to have luck with D3D.
-
@karajorma
I agree to you.
Env mapping for OpenGL? That would be awesome... That is, after Env mapping for D3D is fixed. Really, things need to be fixed.
Do you know how to program for D3D? Cause I sure as hell don't. Bob is probably the only person we have left who does. Quite frankly the team is very tempted to ditch D3D completely as it would make our lives easier. Until Bob comes back and starts working on it or someone joins the team who understands D3D I'd suggest that you consider the entire API depreciated. Hell the only reason I haven't suggested releasing Launcher 5.4 which can't select D3D as a means to hide it is because occassionally someone can't run OpenGL but manages to have luck with D3D.
hmm...
i think....
launcher default API change to OpenGL and (not recommended, unsupported) add to Direct3D label.
ok?
-
Seeing as how the Launcher is being replaced soon anyway it's not worth the effort of updating this one.
Taylor may do that for his new launcher though.
-
To be fair to ShivanSPS his first attempt at a big download had some errors he's hopefully avoided in his new repackage. We simply need to get it up and torrented.
It's also worth pointing out that he isn't a member of the SCP. Simply someone who saw that things could be improved who got off his arse and actually did it. That's what we need more of in the community.
No, no, no... I didnt mean it that way. ShivanSPS's release was absolutely fine when it 'came out of the box'. What I am suggesting is that the problems that were experienced are due to the nature of SCP. I wasn't pointing fingers at anyone, just pointing fingers at a time where certain things were incompatible with many other things. And thinking back, these times happen quite a bit. That package was just unfortunate enough to be made during the peak of one of those times.
Anyway, as to the D3D8 situation, I still believe that it should be fixed, but I agree that it should be abandoned after that. Thanks to Taylor, OpenGL shines in comparison (pun intended). Perhaps leaving it unsupported would be a good idea, whether its fixed or not.
For installers, I have a few ideas of my own, but SCP needs to agree to everything as far as thats concerned. I've been studying Inno Setup and its scripting language, and I think that's best for it. With modular installation, such as installing adveffects.vp for high end video cards, installing FSPort for those who wish to play thru FS1, and installing debug builds. I think this is the most viable option, but the team has to decide exactly how its done. We cant have every Tom, Dick, and Harry making their own package.
-
I have to agree with scuddie here, the project is a bit of a mess. there seem to be more new features than bugfixes, released test builds come along maybe once a month (which would be fine if they were regular and built with consistant featureset and dependancies). Honestly, if this project hopes to survive and really be taken seriously there needs to be a major refocus of the SCP group's talents on to making stable, bulletproof (as much as possible) builds. I come here every day, and every day i wish i had more time so i could contribute if only through organisational changes or bugfixing. I'm glad scuddie brought this topic up, because it focuses on issues i've noticed since i started following the project, but such a topic wouldn't have been tolerated coming from a newbie :sigh:
-
Well, even moders have a life too...
-
For installers, I have a few ideas of my own, but SCP needs to agree to everything as far as thats concerned. I've been studying Inno Setup and its scripting language, and I think that's best for it. With modular installation, such as installing adveffects.vp for high end video cards, installing FSPort for those who wish to play thru FS1, and installing debug builds. I think this is the most viable option, but the team has to decide exactly how its done. We cant have every Tom, Dick, and Harry making their own package.
Taylor is working on a multi-platform installer.
-
And ShivanSpS pulled out the menacing Crushing Fact of Reality +5. The truth shall set them free.
When considering that people do this when they have time to do it, and for their own leisure, the project has managed to survive extreamly well. This kinds of projects do need someone with the overall picture to (at least predent to) coordinate the action. But it cannot be perfect. If our part-time coders would only release totally bugfree, perfect builds, they would be lucky to get out one offical build per year. You just have to accept that a project of this magnitude and complexity will have a few "bumbs", if one does not like that then stick to the retal build. It is true that "casual" gamers may get scared of the amout of effort this game needs to start working. Or they just throw stuff together, barely reading any instructions, and then die away when the first glitch shows its head.
I see things are holding up well. And I don't mind a bit if I have to do some work to get this game running. It is a inadequate price to pay.
Well ofcourse I wouldn't mind if the game would be in a 5MB package. Would install and start itself. And a 5 000 000+ poly digital arm would come out of the monitor and hand me my joystick, open me a beer and shoot SAAA'a at the flies buzzing around my sweaty, greasy carcass of a body.
-
I would've loved to see a 3.6.8 within a couple months ago, I remember it was pretty stable.
-
It's getting better again. A concerted effort from everyone would help. And I don't mean just the coders. Mantis isn't fire and forget. You have to respond if the programmers post a bug note asking you a question.
I tend to find that many bugs have simply dead ended with a programmer asking a question. Those of you who want to see stability in the code should go to Mantis and set it to show the bugs you reported. A few minutes saying whether things are resolved or answering questions would really help the project a great deal.
-
I looked at mantis, and it's a total mess. I see so many reports of crashing when -decals is selected, it makes me wonder if it's efficient at all. Is there a sticky option so that people read what not to report, and what not to do? Also, I see open threads where the OP hasnt responded in 3 months. That's ridiculous. I would say an auto close after 21 days of inactivity.
-
-decal is really working? as far i know is not finished yet....
Anyway, I dont see any "Tom, Dick, and Harry" making their own package, i the only by now :P
When the multi-platform installer is ready will be handy, bu I dont think that the thing going so easy anyway, unless someone make a "Auto-updater" :P
-
Well the cross-platform installer is officially billed as a "Installation Management System". It can be used as an installer, but it's basic function is to manage an installed app. This will mean (at least I hope so, I've got a lot to do and very little time to work on it) that it will show you the installed versions/MODs/TCs, and as long as updates are registered and published for those things, give you a one-click ability to keep up to date with all of them from one app. How well this will work in the end has yet to be determined, the thing is still basically on the drawing board. I had originally wanted for this to be two apps, but later combined the idea into one app, but it may get separated again before going public.
As far as -decal goes, it's very buggy, and has really bad memory issues. Basically it's not supported. Though I guess it's a launcher issue too since it's considered an experimental option by us, but not by the launcher. Easy to fix what the launcher reports it as, just have to remember to make the change.
-
Yeah is a lot better as a Management of an already installed ap... An installer, well I just make one in 30 min which Setup Factory :P, but only supports windows... If there should any way to can make an .exe can be used by any platform... I have a idea, but I not have the enoght knowloged for do anything.... I not even know if viable...
You are making the new the launcher? remember that the launcher should always display the mod.ini of the retail folder when no mod is selected, the problem happends when, you mave the entire folder to a new pc or load a new CVS which no mod selected, when you go to se the MOD tab, the mod.ini of the retail folder is not displayed... :S
-
mod.ini files will be usable with the new launcher, but will officially be depreciated. It's neat, but was never a good long term solution. And the new launcher won't look anything like the current one so don't make any assumptions about that until it's released. It will be much more streamlined, the initial screen having only a few buttons on it, and all options will be sorted into two new screens, general and advanced. A profile selector/creator will also be a seperate option off of the main window and will be where the mod.ini files can be referenced by the code.
The profiles thing is the new mod.ini deal, only far more advanced. You will be able to create a profile for any of your favorite mods, retail, etc. and that profile will include the options that you prefer for that mod and even which exe to use for it. After you have your profiles created (if you even choose to use them) then you will just select the profile from the main screen and hit "Play!". Get done with that game, choose another profile, hit "Play!". No reason to keep changing your active exe or current options.
You will even be able to just create a shortcut for something like Derelict which has this as the command line: "c:\games\freespace2\launcher.exe -profile Derelict -play" and you won't even have to touch the launcher to get all of your preferred settings for Derelict. The game will just start will all of your prefered settings. The -profile option will also be a cmdline option on the regular FS2_Open exe so that you can just call that exe directly with the profile you want and off you go.
-
One point on that. Can we make it so that it doesn't record a default exe if we don't want it to? I don't want to have to update 5 different mods if I download a CVS build :)
-
One point on that. Can we make it so that it doesn't record a default exe if we don't want it to? I don't want to have to update 5 different mods if I download a CVS build :)
Yeah, there is a "<Current>" option for build selection. Basically it just means blank, and the exe from the default profile will be used instead. The default profile is created and maintained basically just like the currently launcher handles it's settings. The main difference is that you'll have to click on the "Settings" button on the main screen to be able to set the build rather than it being right there in front of you.
So basically the "<Current>" profile is what is default. Only when you save the current profile to another name and have it set as active are you not making changes to the "<Current>" profile anymore. You can avoid using the saved profiles feature all together if you want, the active profile, the default one, will just give the same basic behavior as the current launcher.
-
Well, even moders have a life too...
No disrespect, but lack of time really isn't an excuse for bad resource management. Sorting out a definitive and reasonably efficient priority and resource allocation system isn't as hard or time consuming as it sounds, and once done can save uncounted hours and frustrations in the future. even a simple list of priorities and goals (the official term is Project Milestones) to work towards can be amazingly helpful. It can effectively eliminate the "where do i start" and "what now?" questions.
The KISS* principle applies here, if you answer either of the above questions with "check Mantis" many people will take one look at the apparent mess it is (it still won't let me register, so i can't look at it myself) and either put it in the "too hard" basket, or pick some trivial thing they can do quickly. This means you end up with all the little bugs fixed and the big ones remaining. (which from people's conversations here, appears to be a common problem) Assigning priorities to bug reports can alleviate this somewhat, provided it is done to a particular standard, in this kind of situation that standard would probably be based around how many people each bug will affect, ie: if a bug affects everybody and have a major affect on playability and/or core features or design goals, that bug is a "show stopper" and is flagged to be fixed ASAP, before any further public release. If a bug is affecting a particularly cool feature, but not a core feature, that bug is prioritized below show stopper status because the feature could be dropped if it is holding up a critical or otherwise major release. Lower again are bugs that only occur in a specific situation and
only affect a small group of players (ie, features for supporting other mods/campaigns. if the creators want their mod supported they can push for the bug to be upgraded in priority or fix it themselves)
Even before looking at Mantis, having a "shortlist" of features and major bugs to be added/fixed for the next release can provide initial direction and give everyone an idea of where the project is at. As items on the shortlist get ticked off it gives the community a sense of progress, which provides a moral boost and gives vital momentum to the project. there's nothing better for a team to be able to look at a list of goals one day and realize that they are on the home stretch to a milestone. It boosts morale and gets people enthusiastic. Productivity goes up and before you know it, you've hit your target, get a new release, a new list of goals and a team raring to get stuck into them.
*keep it simple, stupid
Wow turned into a long post. Will wait for reaction to this lot before going too far into project management 101. I'm sure some of you have experience managing this kind of project and have some better suggestions. Please no shooting down or flaming anyone's ideas without providing some positive feedback, we all have FS2's future at heart, and negativity only makes things worse.
-
The problem is, it takes new blood to keep a project going and exciting. I see that alot of the original "movers and shakers" of previous years that have created and updated alot of material have either left or have much less time (probably because they're not as interested in Freespace 2 as they once were) This is a shame, but you can only combat this by bringing new people in who get excited by it.
-
*sighs*
*Pulls holy water armed flame thrower from beneath seat*
I personally think that the programmers are doing an excellent job, and if you try a build from a year or two ago you'll realize how much they have done. I don't know of any other game MOD community that has progressed to the point of this one (besides that of Doom 3), and in the end, this progress has been carried on the shoulders of the programmers. If you really care about new blood in the programming department, why don't you try it? And when you get down to it, the thing that really kills a community such as this, is negitivity.
-
I'm not talking about the programmer department (as i get more familiar with the code I will be joining it, spare time depending) because it seems like there are some good programmers involved, but more on the modelling / campaign mission end. We have several "big" campaigns that have been in the works for many years without solid SCP versions released to the public. (Blackwater, Inferno) These campaigns would add a LOT to the appeal of the SCP but they've suffered from not enough people working on them.
I don't think it's a good idea to dissuade people from constructive criticism of the SCP if it can lead to improvements.
-
I dont know why it took me so long to reply to this (perhaps because I have a short attention span), but Taylor has inadvertently brought something else up. The new launcher is going to be re-done and almost unrecognizable... But you're calling it launcher 5.4? Why not launcher 6 if it's gonna be re-done? The same gripe applies to versions of FS2Open. At the current rate of builds, and system, it would take 16 years before we see FS2_Open 4.0. With that, Dec 29, (insert other releases here), Mar 23, and Apr 21 were good CVS builds. Why weren't these 3.x.1, 3.x.2, 3.x.3, etc respectively? As far as I know (which I could be wrong), the first number applies to major version, the second applies to minor version, third to revision version, fourth to build number. So why is it that every six months, FSO is incremented by one revision version? That doesnt make alot of sense.
-
The new launcher is going to be re-done and almost unrecognizable... But you're calling it launcher 5.4? Why not launcher 6 if it's gonna be re-done?
Umm, don't remember saying that the new launcher was going to be versioned as 5.4. At least it wasn't me. Currently I have no intention of keeping with the same launcher versioning. I may end up doing that (though in that case it would be at least 6) if it's ultimately decided that it's the best thing. Honestly though, I don't give a damn about the launcher version at this point. It's currently 0.8, but what it is when publicly released is not a concern for me right now. The new launcher isn't just a launcher, it's a significant upgrade to FS2_Open as well. Profile support has to be integrated into the game, ini file support has (or "had" I should say) to be added for Windows, and the cmdline options had to be completely gutted. The new builds of FS2_Open with those changes will be totally incompatible with the old launcher. The new launcher will support old builds (though the launchers new feature set will be crippled) but the old launcher will basically become junkyard scrap.
The same gripe applies to versions of FS2Open. At the current rate of builds, and system, it would take 16 years before we see FS2_Open 4.0. With that, Dec 29, (insert other releases here), Mar 23, and Apr 21 were good CVS builds. Why weren't these 3.x.1, 3.x.2, 3.x.3, etc respectively? As far as I know (which I could be wrong), the first number applies to major version, the second applies to minor version, third to revision version, fourth to build number. So why is it that every six months, FSO is incremented by one revision version? That doesnt make alot of sense.
The version corresponds to the number and type of changes since the previous build. Though it usually goes x.x.1, x.x.2, it doesn't always. Case in point, the 3.6.5 release which followed the 3.6 release. That's the first one which had the new pilot file formats so it got a larger bump in version to identify it as a larger change. I doubt it will take 16-years to hit 4.0, but I do know for sure that when we do get to 4.0 it will have a ton of new features in it. But we do have to work towards that point, before we have to start that stupid FS2_Open XP, and FS2_Open XP² crap that would happen when we do major versions without true major version feature sets.
-
I think you misunderstood. What I'm saying is that instead of calling the next one 3.6.8, it would be 3.7, the following big release would be 3.8, and all the smaller stable CVS builds would be revision versions. And don't forget, you dont have to worry about reaching 4.0 just yet. It is perfectly viable for 3.11.4 to exist. It's been done many times before ;).
I just think it's easier to give a version than to give a CVS date.
-
who cares what the version numbers are, as long as there ARE new versions....... i'm amazed we're still have a coder like taylor who is adding so much to the game this late...
-
I think you misunderstood. What I'm saying is that instead of calling the next one 3.6.8, it would be 3.7, the following big release would be 3.8, and all the smaller stable CVS builds would be revision versions. And don't forget, you dont have to worry about reaching 4.0 just yet. It is perfectly viable for 3.11.4 to exist. It's been done many times before ;).
I think you're missing something rather subtle. Ask not why the version number isn't 4 yet. Ask why it started at 3 :D