a little DrakX history:

june 1999: 
  i start rewriting redhat's install in perl, partly for the fun of it. I'm
  still working for the army

5 july 1999: 
  i start full time job at mandrakesoft. But we don't have many computers and i
  must share the accounting computer with Merieme who is working half-time. No
  test machine (i test on others box, and destroyed Jacques partitions once),
  guess how it slows things down?

august 1999: 
  at last computers, even test one
  first DrakX version which can install things, very very rough

september 1999: 
  a friend of mine help me 2 weeks on DrakX relayed by Francois

november 1999:
  first released version of DrakX (goldpack). Not really stable yet.

january 2000:
  7.0 is out, with a DrakX quite stable

july 2000: 
  dams starts working on draknet

mid-october -> mid-january 2000: 
  gc rewrites the stage1 to win every kb we can 
  -> size divided by 7 (!) for cdrom


The DrakX team is also doing a lot of other things:
- drakxtools
- urpmi, early MandrakeUpdate, early rpmdrake...
- Mandrake Control Center (new DrakConf)
- packages maintenance/enhancing (esp. ghostscript, 3D-wrappers, lilo)
- fixing core packages to make them installable
- scoring packages, sorting them, flagging them... (compssList, rpmsrate)
- reading/answering cooker and other MLs
- helping non-perl gurus :)

That doesn't give much. Me being the one more working on plain DrakX. It gives 2
people working for 1.5 years. IMO it isn't ``spending an enormous amount of
resources''.
And what do you mean by ``compared to the code base size''??? DrakX is currently
around 28K lines, which is big IMO. You can compare it with linuxconf which is
170K lines. I think the achievement of DrakX is comparable (a 6 times code size
win from dumb C++ to expressive perl is normal imo).


Also it seems like we don't have the same understanding of the word
"maintenance".
DrakX functionalities have evolved *a lot* since the beginning:

- hardware detection, configuration, debugging, work-arounding (multi-kernel installs...)
- making things prettier
  - more bitmaps
  - "advanced" button
  - syslinux graphical boot
- making it work with latest versions of software (eg: switching to rpm4)
- finding out the best way to use rpmlib
- multi-cd
- draknet: configuring every piece of stupid protocols
- diskdrake: raid, loopback, LVM, resizing, checking stupid users entry
- XFdrake: multi-mice, multi-heads, 3D-accel
- porting to axp/sparc/ppc (with Stew's help)
- always more i18n
- keeping things small
  - .cz format
  - moving to .png
  - getFile on demand from mdkinst for ramdisk installs

[...]

> You don't believe the books ? Count yourself; see the man/months spent
> in 'pure' developoment in drakX, and the resources put in mantainance
> of the code base; as far as i know, your count should confirm what the
> books says; more probabily, you will discover that our numbers are
> even worse.

if you count enhancements, adding features... in maintenance, i confirm the
numbers, and find them quite normal.

if you only count bug fixing, the time would be much shorter (around 30%).