Hi!

Samir Saidani <saidani at info.unicaen.fr> wrote:
> goran at krampe.se writes:
> 
> > Hi!
> >
> > Samir Saidani <saidani at info.unicaen.fr> wrote:
> >> I agree with you, but I think we don't need to ask for an agreement :
> >> just replace your initials sd by 'StephaneDucasse'. Then you can use
> >
> > Mmm, and if everyone start changing their initials we need to deal with
> > the fact that there is already lots of code out there with the old
> > initials too. 
> 
> True, that's why I keep the old one and my old squeakmap account to
> tackle this problem. Anyway there is probably initials which

Yes, I have seen several people doing the same. That is good at least.

> correspond to nothing in squeakmap since the initials convention was
> taken before squeakmap.

Note that I migrated all known initials that were written down in the
Squeak Swiki page when I released SM2. Sure, all aren't there though.

> Another way should be to change all my
> initials in the next release by the new "initial" : I will see how to
> do this kind of script. For the older image, I keep the old squeakmap
> account for a while, then delete it after few years, since probably I
> will unable (and don't want) to answer questions of what I did few
> years ago : just read the code and consider that it is anonymous.

There is a problem with that, see below.

> >> Some people wants initials and other wants something more explicit. I
> >> think that the economic way is to choose by default something like
> >> 'DavidSmith' which is more explicit than 'ds' : 'StephaneDucasse'
> >> encodes more information than 'sd', and we don't need to make a
> >> connection to SqueakMap to know the name : so it is economic. 
> >
> > There is no "connection to SqueakMap". You have the SqueakMap instance
> > inside your image - the lookup is just a Dictionary lookup, no network
> > involved.
> 
> Anyway, replace "connection" by "calling instances", this is the same
> : dissipation of energy, when it is not necessary. The initial can
> encode easily this kind of information - the name.

Sure, I agree that having a more readable convention of what we use as
dev initials is fine. But I still think it should be used as the key to
SM, because there we have more information about you, like your email
for example.

> > So let me recap what I think:
> >
> > 1. We need unique ids of some kind. The developer initials has served
> > this purpose until now, and sure, it is not perfectly readable - but it
> > is what we have and we need to think a bit IMHO.
> 
> So let us think a bit now. For me, I see that this is economic to
> encode the name into the "developer initials".

I only agree if we still make sure this is the key to your account on SM
and that we make sure we keep track of old dev initials, again see below
why this is important.

> > 2. If we want to have a more readable field then we seriously need to
> > consider the fact that just renaming our ids to something else will
> > cause "information loss". If I rename myself to "GoranKrampe", then all
> > code with "gk" will be hard to map to me.
> 
> See above, we can also maybe give the possibility to have at least one
> alias for developer initials in squeakmap : keep the old one if you
> want and let us use a readable one.

After chatting yesterday with some people on #squeak several people
wanted the ability to enter multiple "old" dev initials on SM. So fine,
I will add a field for that in which you can enter a space separated
list of old ids. They will still have to be unique, so the old ids are
"forever booked".

> > 3. Adjust the tools so that we stuff not only the initials but also the
> > full name into changesets and MCs. Then we have both and all are happy.
> 
> > And of course, I am not sure how serious problem we have - but it seems
> > less smart if people start renaming their dev initials so that we lose
> > the ability to track all old code...
> 
> I see that I made an error by using an unreadable developer name, so I
> decide to get a new readable name for developer name. For the tracking
> old code issue, personally I never contact a developer for a question
> in one particular method for few particular lines after few
> years/months : it will be silly.

Now, the whole point with being able to track *indefinitely* is a simple
one: code ownership

That code is copyrighted by you and it is vitally important for open
source projects to keep track of where code comes from. Just see the
whole SCO vs Linux shebang.

So no, I don't think it is fine to simply "forget" old dev ids. That is
a bad thing. Hopefully it is now clear why I think we need to "think a
bit" before we start changing dev ids all willy nilly. Ok?

> Cheers,
> Samir

regards, Göran