OK, here it is in writing: I am absolutely serious, "org.gnu" takes more
energy than "gnu". But it's not about typing energy, it's about thinking
energy, the minimization of which is (IMO) one of the principal factors in
designing maintainable and reusable code, *especially* libraries. Having
standards can reduce mental energy as well, but this particular standard has
not helped me personally and does actually clash with my mental model of the
use of libraries. (This is definitely a personal opinion thing.)
It is not the four extra characters that presents the problem to me, but it
is the bizarre naming convention. In my mind I do not associate libraries
with the _domain names_ that created them, why *would* I? And then there's
this weirdness with reversing the domain name (gnu.org -> org.gnu). I do
understand the rationale behind that, but it still goes against the normal
ordering.
In the end, commonly-used names should be descriptive of people's internal
model. That is why software evolved past 'a1' and 'b1', but that is *also*
why you don't use words like "the" in variable names (well, most of the time
anyway). While they don't hurt, they don't help, and you later have to
waste the mental energy throwing out the irrelevancies before you get to the
program's meat.
As to a globally unique identity, I am by no means suggesting that all
packages everywhere do this, but GNU will end up with a pretty large
namespace and is pretty easily recognizable ... I see no problem with its
toolkits "usurping" the higher level name for itself.
gnu is not a higher-level namespace that is likely to be reused by other
corps., and it is not ugly, and does not waste mental energy.
Again, I speak only for myself here. I think it's ugly. I truly do not
know what "most people" think.
--John Keiser
> -----Original Message-----
> From: Thomas J Lukasik [mailto:[EMAIL PROTECTED]]
> Sent: Sunday, September 12, 1999 9:19 PM
> To: John Keiser; Classpat 2
> Subject: Re: org.gnu vs. gnu (was Re: Congratulations)
>
>
> What a ridiculous argument to defend the violation of a
> convention: because
> "it's four less characters to type" (only once at the top of the file no
> less!!) That has all of the merit of someone defending the going back to
> using variable names like 'a1', 'b1' and 'b2' because they are "less
> verbose". I think that the software development world has
> evolved way past
> you guys.
>
> Four 'extra' characters, one time in an import statement, in
> order to obtain
> a globally unique identity, and adhere to a convention accepted
> by the rest
> of the Java development community, is going to be "..a hell of a lot
> (harder)"?!
>
> C'mon, you can't be serious!! Tell me in writing that you're serious.
>
>
> -----Original Message-----
> From: John Keiser <[EMAIL PROTECTED]>
> To: Classpat 2 <[EMAIL PROTECTED]>
> Date: Sunday, September 12, 1999 9:01 PM
> Subject: RE: org.gnu vs. gnu (was Re: Congratulations)
>
>
> >Wow, didn't know this many people were alive and computing on Sundays :)
> >
> >My two cents: gnu.* is a hell of a lot easier to use and think about than
> >org.gnu.*. I still think it looks silly to import com.sun like
> in JavaDoc.
> >But I perfectly understand them putting stuff they *don't* want other
> people
> >to use (or don't care) in com.sun.*.
> >
> >Sun gets to create public interfaces in java.*, why don't we get
> to create
> >public interfaces in gnu.*?
> >
> >Frankly, there are only going to be a few major contributors
> whose toolkits
> >are going to be *extremely* widely used, and GNU is probably one of them.
> >Making it easy to type and use and think about (less mental
> friction) is a
> >good thing.
> >
> >--John Keiser
> >
> >> -----Original Message-----
> >> From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Thomas Down
> >> Sent: Sunday, September 12, 1999 5:58 PM
> >> To: Aaron M. Renn
> >> Cc: [EMAIL PROTECTED]
> >> Subject: org.gnu vs. gnu (was Re: Congratulations)
> >>
> >>
> >> "Aaron M. Renn" wrote:
> >> >
> >> > I'm opposed to org.gnu for the same reason that I hate email
> addresses
> >> > of the form [EMAIL PROTECTED] It purports to be some type of
> >> > globally unique identifier when in fact it is not. In the meantime,
> >> > it requires names that are extremely verbose. At my last company
> >> > we had two Gregory R. Barrett's. One of them ended up a
> >> gregory.r.barrett
> >> > and the other at greg.r.barrett. It's like making your hash table
> >> > really, really big in the hopes you won't have a collision.
> >>
> >> I'm not quite sure that follows... so long as each package has a
> >> defined owner who controls what goes in it, there should never be a
> >> namespace collision using the standard Java system. Okay, code-forks
> >> spoil this a little bit, but I think the principle still holds.
> >> Actually, I'd go so far as to say that one of the really cool things
> >> about Java is that it DOES encourage people to give every class a
> >> globally unique identifier, and this is a Good Thing for efficient code
> >> reuse.
> >>
> >> It's an unfortunate fact that it's hard to have useful unique
> >> identifiers without using some form of central registry. One option
> >> would have been for Sun to run such a registry themselves, but I'm sure
> >> that would have led to all manner of cries of foul play (and they might
> >> well have charged for registrations, which I dare say would have led to
> >> even more complains). Using the DNS as the registry for Java package
> >> names seems a very neat way out of the problem, and while I accept that
> >> the names aren't especially pretty, I don't think they cause any real
> >> problems -- and normally they only affect the package and import
> >> statements at the top of the source files, anyway.
> >>
> >> It's kind-of a pity that there are sun.* packages, but to be fair, some
> >> of these must have existed before the current naming scheme was
> >> developed. And Sun do sometimes set a good example: before
> Swing moved
> >> to its new home in javax, it was always com.sun.swing, and all the Jini
> >> stuff lives in net.jini.
> >>
> >> In reality, having things in gnu.* isn't actually doing anybody any
> >> harm, but I'm not sure it's something that anyone can get moralistic
> >> about, given that Sun are pretty clear as to how things are supposed to
> >> work, and you can't really argue that it's not fair. Of course, if
> >> people asked Sun nicely, they might even approve gnu as an `official'
> >> top level package -- but think carefully before trying that one, it
> >> might just be opening pandora's box.
> >>
> >> Thomas.
> >>
> >>
> >
>
>
