[RULE] do we need a spreadsheet of the software?

Martin Stricker shugal at gmx.de
Mon Feb 24 03:24:20 EET 2003


Eugene Wong wrote:

> I was thinking about how we could improve network installs via ftp &
> http. I thought that it might be valuable for us to try to create a
> spreadsheet of applications from http://www.rule-project.org/en/sw/
> I am hoping that network install programs can download this
> spreadsheet, then use "cut" to parse it, then start downloading the
> appropriate packages. This would allow the program to always have
> access to the most up to date list, without having to update the
> program just for that list.

While I like the idea to enhace network installs, I don't like the way
you want to it. Do not "make a spreadsheet" but make s script that will
retrieve the data *directly*from*the*database* and make it format that
data for your needs. And the needs of the person who wants to install
RULE are *different* from the needs the installer has! *Never* store
data in more than one place (except for backups of course! ;-)) ),
that's what a database is for, so *use* it!

Oh, and never show something that needs to be parsed more than
*absolutely* necessary, because that will increase the hardware
requirements of the installer (and is just unnecessary bloat).

The installer should send a request URL (or POSTdata) to the webserver,
stating several input parameters like available RAM, available harddisk
space, type of CPU, categories of software to install. Then the script
will query the database according to it's input, and return the
appropriate data and formatting:
- The human will want to see name of package, needed disk size,
  recommended or not, and a short description in the chosen language.
  For humans, fixed-width ASCII tables (or maybe easy HTML for display
  through lynx or (e)links) will be best.
- The installer just wants the URL to download and install the RPM,
  of course in the correct order to avoid dependency problems. One
  URL (and nothing else) per line (or several ones if there are
  circular dependencies like some versions of apache and mod_ssl,
  they need to be installed with the same RPM call) is what the
  installer will prefer.

This way the human will have it easy to decide what to install and what
not, and after that the installer will just have to download and install
line after line without any parsing, dependency checking etc.

> If we decide to go ahead & do this, I would be willing to make the
> 1st file. Hopefully, we could be able to have a program or script
> create this automatically for us, eventually. Also, I would like the
> file to be placed in http://www.rule-project.org/en or
> http://www.rule-project.org/

As said above, do *not* create a file (which would always be outdated)
but retrieve the data directly from the database when a request comes
in! This will even make it possible to have the short descriptions in
the user-defined language (they have to be stored in the database, of
course). Don't create a file, not even as a temporary solution, because
those tend to be around for *ages* (one example for that was first
called DOS, then renamed to Windows ;-))) ). Don't waste your time with
a task that will have to be performed over and over again, but invest
your time in doing The Right Thing(TM) and create a script which will do
the job forever.

As said, I *really* like your idea, I just object the way you try to
accomplish it. I'm a database guy for quite some years now, and it
always gives me a small heart attack when I see redundant data or
repeating tasks that are done by anything but a script! ;-))) Thanks for
your *excellent* idea, Eugene!

Best regards,
Martin Stricker
-- 
Homepage: http://www.martin-stricker.de/
Linux Migration Project: http://www.linux-migration.org/
Red Hat Linux 7.3 for low memory: http://www.rule-project.org/
Registered Linux user #210635: http://counter.li.org/


_______________________________________________
Original home page of the RULE project: www.rule-project.org
Original Rule Development Site http://savannah.gnu.org/projects/rule/
Original RULE mailing list: Rule-list at nongnu.org, hosted at http://mail.nongnu.org/mailman/listinfo/rule-list




This full static mirror of the Run Up to Date Linux Everywhere Project mailing list, originally hosted at http://lists.hellug.gr/mailman/listinfo/rule-list, is kept online by Free Software popularizer, researcher and trainer Marco Fioretti. To know how you can support this archive, and Marco's work in general, please click here