Who’s trying to be realistic by attempting to release software with some quality expectations, notably by making the upgrade process seamless and introducing as little bugs as possible?
The problem with that argument is that 5.10.0 did contain regressions.
That the rarity of releases is justified by the high quality standards being adhered to is, unfortunately, merely polite fiction. What it all boils down to is simple:
Shit happens.
No matter how much we’d like for it to be otherwise, shit happens. The release process needs to be reality-based: it needs to deal with the very real fact that shit happens and has in fact already happened. The only realistic way to deal with that is for releases to be easy enough that the shit that does invariably happen can be dealt with soon enough, so that it won’t cause too many people too much trouble.
And the best way to make releases easy is to make them frequently.
And the simplest (though not only) way to ensure that is to put them on a schedule.
That is all.
Whenever I hear someone saying we should have regular releases, I hear we should release when Venus enters Pisces.
There is your misunderstanding, then: what they are really saying is we need a reliable plan for dealing with accidents. If you can make that happen without sticking to scheduled releases, that’s fine too.
Yes: releasing often does not affect the quality of software by itself. But releasing it rarely almost certainly will – negatively. Because shit happens.
I don’t like it either! But that’s reality for you. I doesn’t care what you like. It just is.
Thanks for a good post. I agree.
Yes: releasing often does not make software better by itself. But releasing it rarely almost certainly will. Because shit happens.
Seems like you have a word missing in the vicinity of will. The opinion that releasing software rarely almost certainly will make it better doesn't seem to harmonize with the rest of the post.
Re:Thanks for the post; word missing?
Aristotle on 2009-07-05T16:14:36
Woops; fixed. Thanks!
Yeah, I'm assuming there's a "not" missing there and, based on that assumption, I absolute agree.
I have trouble understanding the arguments against adopting a regular release schedule. The implication is that quality will suffer if releases are made based on the date rather than the readiness of the code - which would be true if the pumpking just bundled and shipped whatever was in the repo at that point in time - but that's not at all what's being proposed. And co-opting DarkPAN to endorse the current release policy just sounds like religion to me: "we're not sure what it is, how much of it there is or even if it exists but we'd better not piss it off."
Re:Absolutely...
chromatic on 2009-07-05T16:46:23
... but that's not at all what's being proposed. That's my biggest objection to the whole debate. I don't particularly like being called a liar, a traitor, an astrologer, a do-nothing blowhard, or a saboteur, but having ideas on which I've expounded voluminously waved away with such mischaracterizing dismissal is unpleasant.
Re:Absolutely...
sigzero on 2009-07-06T00:31:53
I find that when people result to calling others names or shutting down the conversation it means they have no answers themselves and are unwilling to listen.
Re:Absolutely...
Alias on 2009-07-05T17:51:11
> or even if it exists
I think it's fairly clear that it exists.
Re:Absolutely...
AndyArmstrong on 2009-07-05T18:03:22
Indeed. I got carried away by my own analogy.Re:Absolutely...
Aristotle on 2009-07-05T18:04:57
And also that there’s a lot of it. But it doesn’t get any more tangible than that.
And in reality the question is how to balance the needs of the past DarkPAN against the needs of the future DarkPAN. Also, how much past DarkPAN code is affected by the regressions of 5.10.0? It’s not one unknown pushing back; there’s an equally big unknown pushing forward that is rarely ever considered.
Re:Absolutely...
Alias on 2009-07-05T22:28:21
> But it doesn’t get any more tangible than that.
Except for the 50,000,000 lines of it identified here.
Re:Absolutely…
Aristotle on 2009-07-05T23:22:17
That’s the GreyPAN (= open source Perl code that’s not on CPAN).
Re:Absolutely...
cjfields on 2009-07-06T05:46:36
Except for the 50,000,000 lines of it identified here.
I think that's a nice sample of what is out there, GreyPAN-wise, and I will be one of the first to congratulate you if you are able to convince the stalwarts to accept that as a representative sample, and perl5 progresses from there.
However, being in science (primarily) and being human (unfortunately secondarily), I know that there will always be a very vocal few who will refuse to acknowledge such evidence and cling to the precious little bits that support their view. Hopefully with enough support (and maybe persuasiveness) we can progress despite them.
Re:Absolutely...
Alias on 2009-07-06T10:10:00
> However, being in science (primarily) and being human (unfortunately secondarily), I know that there will always be a very vocal few who will refuse to acknowledge such evidence and cling to the precious little bits that support their view.
In the face of actual implementations of proof (or implementations of alternatives that make the original argument moot) people usually come around.
It's happened a number of times.
Re:Absolutely...
cjfields on 2009-07-06T13:04:31
I hope so. The whole mess has led to Raphael resigning.
http://www.nntp.perl.org/group/perl.perl6.meta/2001/02/msg647.html
Now I do. Look at it and blush those of you who are happy at driving out a productive volunteer.
At least silly Simon could eventually apologize before he too crashed and burned.
Do I think the frequent release schedule is helping Parrot ? Not as much as Jonathan is. At the moment the focus is right back on the performance of the VM that Dan was so adamant about. Have those that succeeeded him really made a better Parrot ?
P6 is moving forward because the spec is in an implementable state. Not because Dan was wasting time, or not doing something someone else felt was better PR.
Re:I feel you are wrong.
Aristotle on 2009-07-15T10:24:00
What exactly am I wrong about?
That 5.10.0 contained regressions?
That they have not been fixed in over a year?
That this contradicts the argument in favour of holding back releases on grounds of quality?
That that means it needs to be possible to cut another release soon as a contingency in case of mistakes?
Re:I feel you are wrong.
chromatic on 2009-07-15T16:57:48
What Aristotle said.
Look at it and blush those of you who are happy at driving out a productive volunteer.
Your implication is complete fiction. I invite you to read, for example, everything I've written about the Perl 5 release process. If you can find a single personal attack against any person, committer or not, I will immediately offer a personal apology.
(You will also find that this debate has been lopsided in that respect.)
Do I think the frequent release schedule is helping Parrot ?
Ask other Parrot committers, including Jonathan.
P6 is moving forward because the spec is in an implementable state.
Ask Patrick and Jonathan for their opinions about regular releases.
