Re: SEUL: Observatons

[Luka@pop3.aracnet.com (Peter) <luka@mit.edu>]

winston@atlantic.net wrote:
>    I may be wrong but if SEUL ignores the information already on the
>computer (whatever it is currently using) aren't we in effect shooting
>ourselves in our collective feet and making SEUL even more difficult
>then it has to be.  There are basically two types of systems that we are
>aiming at a pc and a mac (SGI, ALPHA and the rest, at least to my mind
>are not used in a home enviroment.  Maybe I am under a misconception
>here.  From using a pc I know that there is a wealth of information
>pertaining to the system already stored in the computer.  Why take the
>approach of 1) hunting for the information by rummaging through the
>computers memory or 2) asking the user when most of it is already
>there??

yes. getting info from the installed system is useful, so there is
in fact space for discussion here...
i don't think grabbing the info is too easy though.
looking only at the pc, you can expect to be migrating from
dos, win95, win nt (3.5x, or 4.0), or os/2, and other less common systems.
this is not a flame at those os's, but simply a fact:
they all do things differently.  writing anything that interacts with
them, even just to get config info, is going to be difficult.
 * you will have to deal with each os individually.
 * your programs will also have to be up-to-date with whatever info the
current version of each individual os returns, unless the os implements a
good abstraction for this (which some of them may).
 * if seul is dependent on this as its main method of easy-installation,
then it will hit the wall when it meets something not on our top 5 list.
  i may not have been clear about what i meant, previously...
as i said, it is worth trying to write utilities to gather info
like this.  they would definitely be useful!  but i believe the
main installer should operate well independently of their existence.
if one of them exists for the platform a user is migrating from,
then it should be possible to invoke it to gather info and then
pass that into the main installer, which is still written to
run independently of the previous os installation.  this is, once
again, nothing againt those os's, but just necessary to prevent the
amount of code needed for us to support from growing too large.
  considering this, a good point to include in design plans for
the seul installer would be to make sure it can run in such a way
as to make use of previously gathered info about the system
(not supplied during the actual install process).  this could
probably be integrated with the ability to save installer config
data between runs of the installer itself.

> [...]
>It has been suggested that SEUL 'borrow' what is good from various
>Linux distributions so why not do the same with MSDOS (i.e. floppy format).
 in case i was not clear, i am not religiously opposed to the use of
freedos for the installation process.
 what i am opposed to (with logical reason) is using any non-free system as
the main installer, simply because it is not free and not everyone has it.
(note that "it" here is not one single system, even just on pc's - it's
any of the top 5 above, plus others.)  this is not a matter of opinion -
the user is simply legally obligated to pay for out os dependency,
and should not be expected to get great use out of it, since linux is what
they are intending to install here.  one other issue to consider is that by
building our installer around someone else's product, we are governed by
what they do with it.
 so in the case of freedos, cost is not an issue. dependency may be an
issue.  and as always, we should analyze what benefits we gain vs. the
alternative of a traditional linux-based install (which i believe is space
and simplicity - except that the simplicity may not happen if freedos ends
up not dealing well with the routines necessary for linux installation -
this is what i referred to ineloquently as "stability" in my previous
message).  this is a strictly scientific analysis of costs and benefits.
 since this is the primary installer we're talking about, i also realized
that people working on it are likely to be more familiar with coding
for linux than for dos - a possible cost to consider.
 that is all i have to offer.
 i will now leave this idea open for others to discuss... 

-peter luka
----------------------------------------------------------------------------
Simple End User Linux Mailing list
To be removed from this mailing list send a message to majordomo@txcc.net
with the line
unsubscribe seul-project
in the body of the letter.
----------------------------------------------------------------------------