Re: Problem "Stable" is not stable

[Christof Thalhofer <chrisml@xxxxxxxxxxx>]

Am 28.08.24 um 20:46 schrieb Benoît Minisini:

> I mostly agree with the diagnosis, but what should I do in partical terms?

Name the next release 3.20 "New" and 3.19.x "Stable" and so on.

> I think there is a misunderstanding: "stable" is related to the API, not
> to the number of bugs a release has.

Yes I know, and everybody here knows it also. But you can't do anything 
against the fact that everybody else in the world assumes that "stable" 
is mostly bug-free which in case of Gambas is not the reality.

> Moreover, to have less bugs, more testers are needed. And usually,
> testers are often users, and they won't use a version if it is not named
> "stable"! A vicious circle...

I don't think so. I myself try to use every new Gambas subversion and 
revision and then I find the bugs. And there are a couple of others who 
do the same. When you announce the next minor version, I have an average 
of five days until you release "Stable" 3.x.0. That's not enough. 
Sometimes I see and report a bug a month or two months later.

I have not the time to do this (trying and testing) constantly with 
Master. But I can do this with "New".

> As for the state of a stable release, it is good enough for my own need,
> i.e. my jobs where we run dozen of servers with Gambas 24 hours a day.
> But only a fraction of the Gambas components is used, and not the GUI,
> pdf and multimedia components, the harder to test automatically.
>
> But the number of domains handled by the Gambas components is now so
> large that my own use cannot cover them anymore.
>
> A difficult problem...

Noone can predict what will happen if we change the release cycle but we 
can look at examples of other open source products.

Look at the mail of Joe1962 where he writes 'There is also the 
LibreOffice way, which "only" requires maintaining 2 versions'.

Alles Gute

Christof Thalhofer

--
Dies ist keine Signatur