Skip to main content
Topic: "3.90.2" and a note for the Future (Read 8405 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

"3.90.2" and a note for the Future

Because of a handful of people on the dev-list who are "annoyed" by my release of 3.90.2 I'll have to state once again that this is an *UNOFFICIAL RELEASE*.  What the hell does that mean?  It means that its 3.90.1 with my fixes, but that it is not "sanctioned" by the actual LAME project.  Apparently it would be more correct for me to not make a bugfixed release at all than it would for me to call it 3.90.2.  Blah.

I do what I do for you guys, the users.  I saw a problem and I fixed it as soon as possible so that the masses could enjoy a bug fixed version of 3.90.  This however, seems to be causing displeasure to a few vocal bunch.

Because of all this, in the future, releases from this site may not be called LAME at all but instead something else.  I'd like to not step on anyone's toes after all...

HLAME anyone?

"3.90.2" and a note for the Future

Reply #1
"LAYHEM"?

thanks for the hard work for us, dibrom!

brett.

"3.90.2" and a note for the Future

Reply #2
Quote
Because of a handful of people on the dev-list who are "annoyed" by my release of 3.90.2 I'll have to state once again that this is an *UNOFFICIAL RELEASE*. What the hell does that mean? It means that its 3.90.1 with my fixes, but that it is not "sanctioned" by the actual LAME project. Apparently it would be more correct for me to not make a bugfixed release at all than it would for me to call it 3.90.2. Blah.

I do what I do for you guys, the users. I saw a problem and I fixed it as soon as possible so that the masses could enjoy a bug fixed version of 3.90. This however, seems to be causing displeasure to a few vocal bunch. 

Because of all this, in the future, releases from this site may not be called LAME at all but instead something else. I'd like to not step on anyone's toes after all... 

HLAME anyone?


This was an unusual case in which a bug affecting sound quality (on one preset) surfaced in a stable version.  Probably the correct solution is to indicate on the Lame web site that a bug-fix version is in the work for release as soon as possible.  But that in the meantime, "unofficial" releases can be downloaded which fix the problem while the wheels of the lame committee turn.  Normally, an alpha version would have sufficed, but the desire is to fold the fix back into another stable, but fixed version.

I wouldn't really want to see lame forked because of a special situation like this.  In theory, forked versions could lead to confusion if different improvements are made in both LAME and HLAME.  In addition to the name change, it would probably be best to change the version numbering system if improvements in HLAME are not folded back into LAME and vice-versa.

ff123

"3.90.2" and a note for the Future

Reply #3
Hi Dibrom, thanks for all the development work.  I for one don't have any problem with your updates and I dont want to see the LAME development split.  However the pace of advancement has been fairly fast - take a well earned break and let the waters settle.

Thanks again

MGT

"3.90.2" and a note for the Future

Reply #4
Quote
Because of a handful of people on the dev-list...
It seems like there's always a handful of @ssholes trying to ruin a good thing. How about this? BLAME :eek:

"3.90.2" and a note for the Future

Reply #5
Quote
This was an unusual case in which a bug affecting sound quality (on one preset) surfaced in a stable version.  Probably the correct solution is to indicate on the Lame web site that a bug-fix version is in the work for release as soon as possible.  But that in the meantime, "unofficial" releases can be downloaded which fix the problem while the wheels of the lame committee turn.  Normally, an alpha version would have sufficed, but the desire is to fold the fix back into another stable, but fixed version.


I honestly don't see the harm in a 3.90.2 version.  Many people on the lame-dev list however, who I dare say are "out of touch" with the community (as in, most haven't the slightest clue what happens on web based forums even though they know they do exist), do not see this as a necessity.  Instead the "fix" is a proposal for 3.91 which may or may not come out in a timely manner.  Given the current track record for release schedules, I'm not putting my faith in seeing 3.91 released in a rapid fashion.  Since there was no response to my suggestion to fix this situation quickly on the dev list (aside from Mitiok, who of course disagreed with me), I went ahead a published a release here.  3.90.2 should have been an official release.  The fact that it was not, and that it is "condemned" by the LAME team to me is a failure to meet the needs of the users.  In my opinion, this has been a somewhat common practice in the organization (lack of a common goal and the desire to aggressively reach that point) of this project in the recent past.  I'm sorry to say that, but it's just the way I see it.

Quote
I wouldn't really want to see lame forked because of a special situation like this.  In theory, forked versions could lead to confusion if different improvements are made in both LAME and HLAME.  In addition to the name change, it would probably be best to change the version numbering system if improvements in HLAME are not folded back into LAME and vice-versa.


My goal for working on LAME has been basically to improve quality and usability.  At times (and even more now) this seems to be at odds with the goals of the official development team.  I cannot understand why this is, but it creates enough of a disturbance among some people that it may be better to entirely avoid this situation by just doing something outside the scope of the official project.  I certainly agree that this will not be beneficial to many of the users of LAME, but it is also not beneficial to stall progress because of petty disagreements about something as trivial as a version number not being "official".. talk about wasted effort.  If I'm working on something, I want to see a progression and an aggressive attitude towards attaining whatever goal is set.  If that means fixing a bug as soon as possible and bumping the release version, then so be it.  This type of thing should be applauded, not chastised.  The fact that it is not viewed in a such a manner is, to me, both very disappointing and somewhat disturbing and certainly makes the idea of working externally without such "restraints" a very appealing prospect.

All of these things are compounded by the fact that I have been "formally" asked to remove the 3.90.2 compile from my site.  I refuse to do so currently because I do not have time to recompile this version and I refuse to just remove this compile and leave people without a proper solution to the problem.

"3.90.2" and a note for the Future

Reply #6
Damn, that's kind of depressing.  At least you aren't being affected by the negative comments.  Keep up your great work, Dirbrom.

"3.90.2" and a note for the Future

Reply #7
You have to see it that way:

Lamers did much work in Lame for more than just some months - even years!

Now there is a final new one - 3.90!!

At this moment there comes a modified version from Dibrom himself that solves real issues with 3.90, so
this is better than Lame 3.90 from all the Lame developers out there. The name of it is Lame 3.90.2 and is
compiled by Dibrom and only to get from Hydrogen.

How would you feel as real developer of Lame that made all the tunings possible that dibrom uses?

Get the new hydrogenaudio compile - it is better than Lame itself ??


-silence- ( i hope )


But i also feel the need of dibrom to correct faults of its extence manipulations to normal Lame.

The respect to former developements should be decisive to new modifications also.

This all shall got into Lame much quiter than at the moment, and if the frontiers were more sensible
it had worked...

Wombat
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

"3.90.2" and a note for the Future

Reply #8
Quote
Originally posted by Wombat
You have to see it that way:

Lamers did much work in Lame for more than just some months - even years!

Now there is a final new one - 3.90!!

At this moment there comes a modified version from Dibrom himself that solves real issues with 3.90, so
this is better than Lame 3.90 from all the Lame developers out there. The name of it is Lame 3.90.2 and is
compiled by Dibrom and only to get from Hydrogen.

How would you feel as real developer of Lame that made all the tunings possible that dibrom uses?

Get the new hydrogenaudio compile - it is better than Lame itself ??


-silence- ( i hope )


The issue here is that ego, once again, should not have anything to do with this.  Progress should be had (bug fixes released as rapidly as possible) and everything else should be set aside.  I did attempt to get a 3.90.2 version "officially" released, meaning that I did try to work with the developers on this before doing it myself, but I did not receive a response fast enough (except for one opposing the idea) so I just went ahead with it.  Nothing is stopping this modification from becoming official except for the "wishes" of a few people.  They don't think it's the "proper" thing to make a 3.90.2 release, so the only option is that  "officially" we have to wait for them to release 3.91.  Who the hell knows when that's going to happen.

As for developing with LAME, I'm certainly not doing anything as complex as what Robert or Naoki has done, but I'm just as much of a "real" developer as most of the other people listed on the LAME page, and I've done more than most even.  One of the most vocal people complaining about my release, to my knowledge has never made a modification to the code itself at all, aside from some small tweaks to the Makefiles.....


Quote
But i also feel the need of dibrom to correct faults of its extence manipulations to normal Lame.

The respect to former developements should be decisive to new modifications also.


Sure.  All my modifications to the code are completely not enabled by defaulted.  I go out of my way to make sure I don't touch the "standard" behavior of LAME since that would probably cause a massive uproar itself.

Also, new modifications I submit and my handling of developments are second to older conventions already in place, but only when they are sensible.  Withholding a bug fix simply because some people don't like the "officialness" of 3.90.2 is not sensible.

"3.90.2" and a note for the Future

Reply #9
Quote
Originally posted by Dibrom
Because of a handful of people on the dev-list who are "annoyed" by my release of 3.90.2 Apparently it would be more correct for me to not make a bugfixed release at all than it would for me to call it 3.90.2.  Blah.

I know it's a bit cheap to say now, but there should have been a Beta, a "stable" version where development and tuning would have been frozen for a month or so, and where only bug-fixing would be allowed to go into the next stable. Testing be done in separate alpha's (just like your rev 9 & 10).
Quote
Because of all this, in the future, releases from this site may not be called LAME at all but instead something else.
HLAME anyone?

I and the other people who read this forum will have no objection to you releasing a 3.90.x but I must admid that is a bit messy and confusing for the "just gimme mp3's" users out there. Combined effort (like LAME development) is maybe slower now and then but the result can be bigger than from one person continueing on his own.

I suggest stick with dm-391revX naming as you did. We'll know what's going on for sure.

Whatever you choose to do, you can't go wrong here as long as keep your ear on the quality
--
Ge Someone
In theory, there is no difference between theory and practice. In practice there is.

"3.90.2" and a note for the Future

Reply #10
Well, in my opinion the bug fixes were the most important thing here. However, 3.90.2 was a bit poor choise of name.. In principle what would become if several people released different (although unofficial) compiles like 3.90.3, 3.90.4 which would include different things. I don't think there is/was any danger at this time for that to happen, but the naming and releases should be either official or clearly different. In my opinion 3.90.2 should have been named rev11 or something else.

Regarding Wombat's message: I don't think that 3.90.2 takes away anything from the Lame devs. .2 is just an addition by one dev, so I don't understand Wombat's sentence "Get the new hydrogenaudio compile - it is better than Lame itself ??". 3.90.2 is of course also Lame with an additional small bug fix to the Dibrom's own code section. It doesn't diminish the value of 3.90 or show disrespect towards other developers regarding the code, it's the same code as 3.90 with a small addition.
Only thing questionable is that since there wasn't going to be official 3.90.2, Dibrom made one himself, and although marking it unofficial, it wasn't still in my opinion the perfect solution. However, this issue is quite minor (and the quality fix was more important issue anyway), and I believe with better communication between certain people this dispute would have been totally avoided.

The point is: one developer can't just release unoffical but "official looking" releases by himself (The naming looks totally official unless you more closely read this site or check the file_id.diz). The problem here is more principle matter than actual problem. Anyway, this was a small mistake which should be acknowledged by Dibrom, and then this should be forgotten and move on with more important issues.

Still, like Dibrom says nothing is stopping the unoffical 3.90.2 version becoming official, except few people who think the quality fix (or Dibrom's work in general) is not important enough. But unfortunately, if other Lame devs think so, then no can do. It's an annoyance but there's nothing you can really do, unless other Lame devs can be convinced otherwise.
Juha Laaksonheimo

"3.90.2" and a note for the Future

Reply #11
In future, you may want to call your enhanced versions something like lame-3.90-dibrom1, to emphasise that this is not an official LAME release. Note however, that what is 'official' in free software has much less relevance than what is 'official' in proprietary software (see the history of the gcc/egcs fork, for example). Given that LAME's licence is the LGPL, then you are free to fork the code whenever you want, and call it what ever you like (unlike something like the Artistic license, which stops you from calling a fork by the same name as the original).

"3.90.2" and a note for the Future

Reply #12
I'm afraid I have to agree with the members of the LAME development team who disagreed with your choice of names for this release. Whether you respect the quality of their decisions or not, it's the responsibility of the LAME coordinators to cut mainline tarballs from the CVS tree, and I don't think it serves the community to condone the distribution of versions that appear to be mainline, but actually are CVS snapshots or have custom modifications. I am aware that you modified the printed version output at program execution time, and placed a notice inside file_id.diz, but unfortunately I don't think that's enough. There are a plethora of custom-compiled and modified LAME versions available on the web, but they're all clearly marked as such in the tarball filename, and I don't see anyone complaining about them.

I don't think there's anything wrong with the CONTENTS of this particular release -- far from it, in fact. I'm glad that you noticed the bug in 'fast standard' and moved so quickly to correct it and make people aware of the situation. However, there are a lot of people out there who are NOT as intelligent as you, and I'd hate for them to think that it was acceptable to make well-meaning but incorrect changes to the codebase and release their versions as 3.90.2 or 3.90.3 or, God forbid, 3.91. I remember very clearly back in my DOS days, people would deliberately release virus-infected "new versions" of PKZIP with incremented serial numbers to trick people into infecting their computers. I'm certainly not claiming that your release is malicious or going to harm people's computers, but that situation caused no end of headaches and wasted time for both PKWare and the users of PKZIP, as they had to determine which releases could be trusted. I think it's important for it to be made clear that this practice isn't acceptable, regardless of who you are and how good your intentions are.

I'm no apologist for LAME; personally I'm holding my breath patiently for the next release of Ogg Vorbis.  I think the project coordinators have made some silly faux pass recently, especially re: the 3.90 & 3.90.1 debacle. A) Why wasn't a final compile from scratch on a clean machine attempted before releasing the tarball? B) They should have bumped the version number in the compile to reflect the tarball version -- if I downloaded 3.90.1 and it identified itself as 3.90, I'd wonder if the tarball I had gotten was corrupted or a bad release.

All of this said, I want you to know that your work on improving LAME is GREATLY appreciated. I only started reading this website & forums a few days ago, when an idle pass through the r3mix forums informed me about the new --alt-preset options and this website. I was thrilled to see that someone was doing new and interesting things with LAME quality tuning, more than just parroting the opaque --r3mix setting as "the best ever", or providing endless commandline options that I really had no chance in hell of understanding (--ath-type ???).

While you certainly could fork LAME off, I'd hate to see that happen for reasons of personality alone. Far too many open source projects have wasted countless hours and effort in code duplication simply because two people couldn't interact civilly. While that sometimes in the end does provide better quality (witness egcs & gcc), I feel in most cases it just slows everyone down as the existing development team splits, and people spend time competing with each other for reasons of spite. Everyone here at hydrogenaudio clearly knows who is responsible for the new tunings; don't let one or two naysayers in the crowd put you on the defensive, no matter who they are or think they are. Heck, half of the version changelog for 3.90 is detailing your --alt-preset work, and I think that tells everyone all they need to know.

Flame away if you feel the need; I won't take it personally. Just understand that everything that is said in here is said with the greatest respect for all of the great work you have done, and in the interest of continuing to make the best possible mp3 encoder available to everyone. I am greatly looking forward to testing Ogg Vorbis 1.0 versus your latest-and-greatest LAME tunings.

"3.90.2" and a note for the Future

Reply #13
Yes, 3.90.2 may have been a poor choice of a name, but I did it for a couple of reasons:

1.  3.90 had just been released.  It would not go over well for me to recommend my new custom compile, or some CVS snapshot over the official "stable" just because of some bug.  Especially where advocacy is concerned, trying to get users to upgrade, it looks especially bad to have to do this.  "3.90 is just released after a year and a half!  Oh, but it's got a bug.  Make sure you get Dibrom's unofficial compile".

2.  I honestly expected this effort to be backed up by the LAME dev in some form or another.  As I said before, and as JohnV said, 3.90.2 should have been an official release, and there is still nothing stopping it from becoming so except for a few people who are hung up on some trivialities.  The next "best" bet we have to have this problem handled "officially" is 3.91.  Who knows when that's going to come out.  I'm certainly not going to tell everyone to wait for that release though, and it's equally unacceptable to tell everyone just to not use --alt-preset fast standard because 3.90 is broken.

3.  I am actually an official developer on the project.  Yes, it would be bad if everyone just started releasing compiles and bumping version numbers, but I'm not exactly everyone.  I participate directly in the project and discussion of it, and I have access to CVS.  I'm also listed on the developers pages.  So, my decision doesn't reflect that of the project maintainers', but it's also not just some completely rogue development either.  Again, 3.90.2 should have been official.

Anyway..  I think I've decided I'm not going to remove this compile until there is another "official" (Non-CVS) version available.  I don't really have time to recompile it at the moment anyway, and probably won't for at least a week or so.

 
SimplePortal 1.0.0 RC1 © 2008-2018