[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Goals for alpha release



I'd like to get an alpha release of d-i out the door soonish.  Most
pieces are there already, but some few crucial pieces are still missing.

- ftp priorities need to be adjusted.  They are used by anna for
  deciding what modules to install.

- cdrom installs, checking whether it actually works at all.  Fix
  bugs, test.

- mkfs/mkswap, mkswap-udeb is not in the archive, mkfs is currently
  called by hand.  I have code for doing this which I just need to
  test

- testing.  The whole installer needs to be better tested.. I can't
  test my own code well enough, so we need somebody to actually run
  through this code and try to break it before we let it loose on our
  users.  Test cd installs, test net installs, test various partition
  setups, etc.

- fix lilo-installer.  lilo-installer doesn't do initrds ATM.  That
  means booting is broken.

Anything which creeps up on us and is obviously broken and must be
fixed will also have to be fixed before we can stick the «alpha» label
on it.

A few more fluffy goals for beta release, as aj put it when I talked
with him tonight: «well, i'd make my alpha release goal be "<something>
works", my beta be "<something> works well".»

So, my fluffylist is somewhat like:

- slang, other frontends, gtkfb?  Some work has been done, but nothing
  solid has come from it yet.  We need to do something clever if we
  are going to have X-based installs.  X udebs would be stretching the
  concept a bit far.

- automated installs.  I have begun hacking on cdebconf to support
  stacking.  Once that is in place we need to have a http backend or
  something and this might actually fly.

- porting.  We need the ports going soon.  BenC has said he is going
  to work on porting to SPARC in the next week or so, the powerpc port
  seems to have stalled, I am not sure about the alpha port status is,
  nor most of the others.  That is, I assume they are non-existent
  until they prove otherwise.

- testing drop-ins of components, like evms.  d-i is modular. I think
  it should be possible to replace the normal fdisk and mkfs routines
  just by dropping in the right evms udeb and selecting that.  We need
  to check that it works and fix any bugs there.

- reconfigure arbitary package.  Some packages, like choose-mirror use
  debconf when configured, but they are impossible to reconfigure.
  Having a method to reconfigure them would be useful.  Actually, it
  might be so useful and easy to implement that I want it for alpha
  release.

Anyhow, this is probably not a complete list at all, just me trying to
organize my thoughts so we can actually get a working installer out
the door in the not-too-distant future.

Cheers, and thanks for the good work done so far,

-- 
Tollef Fog Heen                                                        ,''`.
UNIX is user friendly, it's just picky about who its friends are      : :' :
                                                                      `. `' 
                                                                        `-  



Reply to: