Re: current

On Tue, 2002-08-20 at 05:27, Joe Radinger wrote:
> 
> 
> hello,
> 
> thanks for that wonderful piece of code (i use 1.0.5 since 2 weeks).
> 
> 3 questions/comments:
> 
> a) i seem to be not able to use current as a server on redhat 6.2. is
> this ok? maybe you should stress that more, if there are minimium
> redhat-version requirements. what are the problems with 6.2? maybe i
> could provide some help in backporting?

Due to rpm and xmlrpc versions, I bet Red Hat Linux 7.2 is the earliest
that will work at this point for the _server_. 

It should work fine with older clients, but the server itself needs to
be 7.2 (perhaps 7.1....)

With The Next Version coming fairly soon, I probably wouldn't be that
interested in a back port at this point. If 6.2 clients don't work for
some reason, that I would take interest in.

> 
> b) maybe someone pointed that out already, but looking at anaconda's
> source could be useful in creating channels. i create heavily modified
> installation cds, based on rh 6.2 (now we start using rh 7.3) and found
> that extremely useful (the scripts for generating cds where undocumented
> and nonworking for 6.2; but anaconda from 7.2 has improved a lot
> [despite being almost undocumented again]).

We know how the channels are constructed. At this point the big hang up
is a data storage mechanism powerful enough to deal with them. The
shelve based (python objects in bsddb files) just can't do it.

> 
> c) as you ask for opinions on future releases (maybe this has been
> discussed already): i would prefer a "small" installation WITHOUT the
> need for postgres and apache, maybe there could be such a beast with

1.0.x works that way, but the limitations of it are pretty severe. I
have located a project which I think will let us eliminate the
requirement for postgresql if all you want is a simple server, but at
this point, the apache part works too well. 

I fully sympathize - I didn't want those requirements either (which is
why I'm trying to use sqlite for us if the user doesn't want postgresql)

I'm not familiar with caudium - does it have an equivalent to
mod_python? I almost certainly wouldn't develop for it, but I'd
certainly take patches.

> both ways? we use on our sites no apache, but caudium as webserver, and
> we dont want to have another one to administer. same applies onto
> postgres (we use mysql). 

I'm fully sympathetic about requiring apache, I just don't see any other
way forward, with any kind of performance.

-- 
Hunter Matthews                          Unix / Network Administrator
Office: BioScience 145/244               Duke Univ. Biology Department
Key: F0F88438 / FFB5 34C0 B350 99A4 BB02  9779 A5DB 8B09 F0F8 8438
Never take candy from strangers. Especially on the internet.