[plug] Controversial comparison of distros?

sscott at iinet.net.au sscott at iinet.net.au
Tue Oct 28 15:35:17 WST 2003


Quoting Cameron Patrick <cameron at patrick.wattle.id.au>:

> You're implying that there is a clear-cut distinction between core and
> non-core ("3rd party") applications and that this distinction is in some
> way meaningful in this debate.

yes, I am :) The distinction being that 'core' apps can be installed any way 
the distro vendor wishes (hell, tar up a filesystem contain all the basic 
stuff... hey, dont they already do that? ;)), whereas 'non-core' applications 
should have a standard and well defined way of being installed. 

> Exactly.  And as such, they make different packaging decisions based on
> the appropriateness for a particular target audience or purpose.  

My point is that the choice of packaging would not materially affect this and 
is therefore nothing more than an added burden to package maintainers. How 
would debian's 'appropriateness' change if the package format was changed from 
deb to rpm? 

If they rewrote apt/dpkg to use rpm instead of deb, would most people even 
notice?

> Ahem.  Stop it with the stereotypes, please.  Irrelevent ad hominem
> attacks are not appreciated.

Point being (for the humour impaired), each distribution has a target audience. 
Id say they were pretty clearly defined.

> 
> | If they all use RPM, would it really matter? If they all had a similar if
> not 
> | identical directory structure, would it really matter?
> 
> Possibly, although some alterations/improvements to RPM may be
> necessary.  (Debian packages, for instance, are allowed to contain
> interactive configuration scripts that run before the package is
> unpacked.  RPM doesn't support scripts that run before a package is
> unpacked, and mandates that its postinst scripts are non-interactive.
> Whether or not debconf is a Good Thing is another matter entirely.)

The rpm format is not set in stone. Fix it.

> 
> | RH could continue tuning their distro for Oracle, Mandrake could
> | continue packaging unstable X apps which give everyone the shits,
> | Debian could continue with their marvellous tradition of testing
> | everything for 14 years and having ultra-perfect dependency trees, and
> | slackware could continue to be ignored by anyone except those people
> | who really should be running one of the BSDs.
> [...]

ugh. more humour. I must learn to stop trying to be funny in the face of blind 
zealotry. Ill add a note to my zire, which I now use since my linux based 
handheld proved to be next to useless. :)

> | 3rd party software (including open and proprietary) would have an easier
> time 
> | packaging and distributing their software. General end users wouldnt have
> to 
> | worry about which distro they were running.
> 
> Surely the above two paragraph are contradictory?  If distros are
> different in some way then end users /will/ have to worry what
> distribution they're using, and people distributing software in binary
> form /will/ have to care about the differences between distributions.

Not true. Youd have to agree, that just about every Linux distribution in 
existance would be capable of running just about every linux app in existance, 
with varying levels of effort. Linux is linux. Just because I have to install 
90 packages of dependencies on slackware vs 2 on mandrake to get DivX's playing 
shows that they are chasing completely different markets. But it is still 
possible on both systems.

> 
> | As soon as you start talking about deb vs rpm, you scare the poor
> | dears.
> 
> ("Dear-caught-in-headlights" reaction? :-P)

I could relate a story about my father's attempts at getting DVDs to play by 
installing a source rpm he downloaded, but I wont ;)

> 
> One could argue that Debian requires deb packages instead of rpm
> packages for the same reason that Red Hat requires rpm packages instead
> of self-installing exe files.  Sure, the differences between Debian and
> Red Hat may be lesser in magnitude than those between Red Hat and
> Windows, but the principle of different operating systems generally
> requiring different versions of software packages remains sound.

Youve lost me here, Im not too bright today. Can you elaborate?


> Finding something that works (or is "good enough") is easy.  But if
> there are other things that work /better/ for a certain purpose, should
> we not be working to popularise them rather than marginalise them as
> "unpopular", and/or denigrate their users as "comp sci students with too
> much time"?  (Me, I'm a maths student with too much time! :-P)

My point is that there are better things to worry about (ie, your maths 
degree!) than the fact that debs have pre-unpack triggers and rpm doesnt when 
rpm seems to work just fine regardless.



_______________________________________________
plug mailing list
plug at plug.linux.org.au
http://mail.plug.linux.org.au/cgi-bin/mailman/listinfo/plug


More information about the plug mailing list