On Tue, Apr 24, 2001 at 02:47:49PM -0700, burtonator wrote:
> Tavin Cole <[EMAIL PROTECTED]> writes:
>
> <snip>
>
> > Ah, but those standards are too bloated and unwieldy for the unique nature of
> > our project! Sure they look attractive but that's because all the bloat is just
> > hidden away in libraries. Naming the base package "Freenet" has cut out hundreds
> > of megabytes of "com.gar.gar.gar.." bloat..
>
> No. Just name it 'freenet' instead of 'Freenet'. There are some strong
> technical reasons why you should have packages as lowercase:
>
>
> A lot of Java developers (I am not one of them) will look at the "Freenet"
> package and instantly judge its source code quality just from the fact that it
> violates the bible - The Java Language Specification:
I tend to agree with Tavin that we simple don't need developers that
superficial. If we are found worthy of passing the scrutiny of the
directory name test, we will fail whatever analism comes next. I have
found that we have had no shortage of people ready to check out the source
and send posts to this list explaining how brilliant they are and how they
will be taking over now - but none of them have ever produced much of
anything.
The reason we have the package name that we do is that is what it was -
and it is a real bitch to move whole modules in CVS, especially if you
don't have access to the filesystem on the server and know exactly what
you are doing (I don't think we have either).
If somebody wants to do the work, I guess we could move it, but I'm very
dubious about using a DNS name. I'm not very fond of DNS, and I don't
recognize that the ICANN has any sovereignty over naming of anything - as
far as I'm concerned we might as well use Napster user names or ICQ UIDs
to identity the packages (here is an idea - lets use an SSK subspace).
Getting java into bed with Verisign is right up Sun's alley, but not ours.
The point with the directory structures you describe escapes me as well. I
mean, with the java system of trying to achieve unique naming of packages
and specifying the root from a system var seems to imply that one should
be the structuring the java classes above a single (or a couple, one
global and one user maybe) class root. But with these directory structures
where the source is down a bunch of directories from any sort of common
root, every single java program has make and run scripts that reset the
classpath explicitly - which is closer to the DOS cd cd cd cd cd run
version then anything that makes sys vars and unique naming matter. I like
the fact that I have the Freenet module checked out into right into the
class tree - not in Freenet/src/main/fred/foo/bar/org/freenet/fred, and I
want to keep the ability to do ctrl-x-s-c-v-c without having to worry
about setting vars explicitly for Freenet (and I wish other projects
would do so as well)...
As far as other people loading Freenet classes, I would recommend they
copy the code directly instead, our interfaces change all the time (which
I'm sure those hypothetical people who are definitely not the Burbinator
himself would go into a type 4 seizure over as well).
<>
> When you start playing with reflection and dynamic class loading having a
> standard naming really helps make things clear. I think Freenet should benefit
> form this :)
The illusion of unique naming does not provide any security that it is
your own code you are loading dynamically. I'm sure that Sun provide a
code signing system to do so - and I'm sure it is equally in bed with
verisign and equally unfit for us...
(btw, Mr. Bad - I've removed all the dynamic loading in experimental.)
> > > If the Freenet project wants to attract other developers it is important that we
> > > adopt some standards.
> >
> > Hmm.. that's an interesting concept. I think we should adopt some orphans
> > instead, then train them to be SLAVE developers!
>
> ... :)
>
> - --
> Kevin A. Burton ( [EMAIL PROTECTED], [EMAIL PROTECTED], [EMAIL PROTECTED] )
> Cell: 408-910-6145 URL: http://relativity.yi.org ICQ: 73488596
>
> In order to see the fish you must watch the water.
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.0.4 (GNU/Linux)
> Comment: Get my public key at: http://relativity.yi.org/pgpkey.txt
>
> iD8DBQE65fSFAwM6xb2dfE0RAt6eAJ9hAPNjRHTk8lKQk0jKfgXrnUr7KgCgnMOW
> S3amDq/3OmGuq5cS0FKEh28=
> =Oejo
> -----END PGP SIGNATURE-----
>
>
> _______________________________________________
> Devl mailing list
> [EMAIL PROTECTED]
> http://lists.freenetproject.org/mailman/listinfo/devl
--
'DeCSS would be fine. Where is it?'
'Here,' Montag touched his head.
'Ah,' Granger smiled and nodded.
Oskar Sandberg
[EMAIL PROTECTED]
_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://lists.freenetproject.org/mailman/listinfo/devl
