Re: [binarycloud-dev] Various

[alex b. <en...@tu...>]

> So, after reading the r2 "specification", this really
> seemed right.  So we are waiting.  I wish to compliment
> you on your very impressive concept and, so far, what
> is in the CVS that we are now playing with.  I suppose
> our only reservation is our worry that binarycloud
> reaches that "critical mass" where basic modules are
> going to be contributed, and "ported" from some of the
> other systems.  We are certainly going to be doing this,
> from Great Circle/Lat/Long/ZIP modules to maybe even
> a dynamic report generator for MySQL.  It is hoped
> that the module collection will allow, eventually,
> slapping together a basic intranet quickly.

cool.

> Just a couple of questions:
>
> 1.  We notice that you use bare variables in the hashes.
> We used to, by accident, when we left quotes off.
> I think bare variables are legal in Perl, but
> the question is -- IS THIS ON PURPOSE?  Maybe
> 18 months ago I asked the PHP list if bare
> variables are "supported" since there was nothing
> in the manual to suggest that they were.  Rasmus
> replied that because of the mechanization of the
> symbol table, or some such, THEY WORK AS A SIDE
> EFFECT, but that "this behaviour is not guaranteed
> in the future".  Upshot, don't use bare variables
> if you don't want your code to blow up in 2003.

This is a result of me not synching CVS with my core repository. I am
waiting to do that when I have a make system that I think is pretty much
final.

I think it would be _really_ _really_ confusing for people to deal with a
moving structure re: the make system.

Re bare vars in hashes, no that's laziness that I fixed a while ago :)

I have copies of everything in CVS that use no bare vars for array keys, as
that is bad style :)
If you turn up error reporting high enough, it will tell you :)

> 2.  Why do you return true from the modules->Init()
> and modules->Output?  Is this just good, knee-jerk
> practice right now, or is there some kind of
> "recovery" mechanism anticipated, say, in the
> future when the returns will be tested.

No, knee-jerk.
It may (read: probably) go away, actually, because as yet I cannot think of
anything that would "govern" Init. I haven't put a great deal of thinking
into non-Page+Modules use of bc, which is why those returns are there...
i.e. I haven't decided how to have people use components of core like
auth/sess/etc, without using Page.

> 3. [Frivolous]  Do you want an email pointing out simple
> spelling errors/typos in content and comments?

Not yet, because 90% of them are probably gone.
After I release r2/make, yes.

I have done some major renaming + structural changes because of the make
system, that included applying proper r2 headers to all files.

If you have any typos in the core class method comments, send 'em :)

> So, we again wish to compliment you, and encourage you.
> It is an impressive piece of work.

Thank you, very much. It's good to hear and it is certainly encouraging.

The make system is opening up so many possibilties, it will be _very_
exciting to release it :)

I realize it has been a bit quietly, but I want to make sure that the next
pre release which contains make is as close to "final" as possible... so
people can start to use r2 for building actual applications.

ohhh, and you haven't seen import, have you? :)

> Regards,
>
> --
> Justin Farnsworth - Technical Director
> Eye Integrated Communications
> 321 South Evans - Suite 203
> Greenville, NC 27858 | Tel: (252) 353-0722

:)

_alex