Installer status (was: Re: [squeak-dev] [ExperimentalCoreRelease] PharoCore-1.0-10508rc2 vs SL3dot11-9499-alpha

Ken G. Brown kbrown at
Sat Mar 13 16:40:10 UTC 2010

At 3:39 PM +0100 3/13/10, Levente Uzonyi apparently wrote:
>On Fri, 12 Mar 2010, Ken G. Brown wrote:
>>At 3:51 AM +0100 3/13/10, Levente Uzonyi apparently wrote:
>>>On Fri, 12 Mar 2010, Ken G. Brown wrote:
>>>>Installer is being actively maintained externally. There were recently some bug fixes from someone on the Pharo list. Installer fixes in trunk have NOT to my knowledge been put in the Installer repo = lose, lose.
>>>Who is the maintainer? Can we cooperate with him?
>>Is public commit as far as I know.
>So there is no maintainer.

Does 'trunk' have a maintainer?
Seems to me that modern software development methodologies advocate that source does not have the author in the source code.
It is felt source code should not have an explicit owner, having an owner is thought to be detrimental to the group development process. People are afraid to do fixes then.
The community becomes the owner and anyone then feels empowered to do maintenance.

>>>Why are the commit messages empty from Installer-Core-nm.335 to Installer-Core-nm.349?
>>I don't know.
>>>From what I understand 3.10.2 was released with an old version of Installer.
>>I think the latest version of Installer and related was installed by installing LPF, but I believe then you would then have the latest MC1.5/1.6 too.
>>Trunk seems to have started with ancient Installer and MC.
>IIRC Matthew recently loaded the then-latest Installer-Core to the Trunk.

Then Matthew is a maintainer since as I recall he gathered together the changes made by others to Installer in trunk and not posted properly to the Installer repo.
So is the fellow from the Pharo list who made some bug fixes to Installer which you probably don't have in trunk.
And having to root around to find fixes to Installer caused extra work for Matthew, would have been far easier for the person making the changes to post back to the Installer repo when the changes were made.

>>>The changes are not lost, they can be merged them back to the squeaksource repository. Feel free to do it if you think they are useful.

Here's the way I would look at it if I were wanting to use Installer (which I do, I think Installer is great).

Do I want to use Installer? Yes/no.
If Yes,
Is Installer an externally managed package in its own repo? Yes/no.

If yes, load Installer from the repo.
Do I need to fix some things in Installer? Yes/no
If yes, do the fixes, test them adequately, when happy post, to the Installer repo.

I would not expect some magical someone else maintainer to search around everywhere every day, eg trunk, or Pharo, etc. for any good changes to Installer, and then figure out how to grab the changes, and do the post to the externally managed Installer package repo. That makes no sense.

>>Those that are working on the package would obviously then want to commit to the repo to keep the repo up to date.
>>>Btw the repository is in a bad shape:
>>>"The currently blessed version of package 'Installer-Core' is Installer-Core-kph.92.mcz"
>>>And the repo is full of unrelated packages.
>>No doubt the packages are related in some way.
>You mean that CollectionsTests, DrGeoII, KernelTests, Split-Join and Universes should be in this repository?

I don't know the reasoning. Maybe they were put there by some simple finger trouble mistake or glitch in ancient versions of MC which people insist on continuing to use.
Actually, if I go to the Installer project on squeaksource when logged in, and click 'Versions' I don't seem to see the ones you are talking about except Universes which has a comment saying "- Added Installers for mcd, mcm file types" so it appears related.

Ken G. Brown

>>Sounds like some things could be cleaned up.
>>Ken G. Brown
>>>>Ken G. Brown

More information about the Squeak-dev mailing list