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

various anaconda observations



I spent the better part of the weekend pulling apart RH9.0 anaconda and 
I jotted down some notes and I'm wondering if anyone else has taken a look 
at some of these areas.

.  Why no graphical url/ftp installation?

   I noticed that there still is no way of doing a graphical ftp or http 
based install.  The netstg1.img file has been renamed netstg2.img (which 
makes a lot more sense) but there still is no provision for doing a 
graphical network install except via nfs (and I guess this is since 
it uses stage2.img).  There shouldn't be that many technical issues about 
getting this to work correctly, so did it just fall off the radar at Red 
Hat?

. Crosscompilation?

  I've been trying to get the loader to cross compile from RH7.2 for 
RH9.0, but the Makefiles are a bit of a mess.  Has anyone else attempted 
to do this?  For the most part I've just been unarchiving each of the 
-devel-* packages which I need and building a RH9.0 cross compilation tree 
(although I'm keeping each of the packages seperate and using explicit -I 
and -L directives).  So far the only real hang-ups I've had that are the 
gettext and bogl binaries are dynamically linked against the wrong glibc 
so I pulled those apart and I was going to statically link them and put 
them back in but haven't finished yet.  Oh, that and I'm still 
building loader against libc and not dietlibc until I get everything working.  
After a lot of Makefile-foo I ended up getting isys to compile with 
dietlibc correctly and tested it out in a VM running 9.0 and it seemed to 
work fine, so it'll probably only take me a bit longer to get loader 
working as well (hopefully).

. Boot.iso/boot.img/bootnet.img/driver disks

I noticed that the boot.img/bootnet.img files were consolidated into one 
file and there are now a couple of driver disk images which I guess 
provide all of the network drivers.  I haven't ripped apart the new boot 
image to see what drivers are in it by default, but I imagine I'll get 
there in the next couple days.  I did like the boot.iso idea as I guess a 
lot more people have CD-R's these days so it probably makes a lot more 
sense than installing off of the floppies.  Did the driver disks end up 
being the same as the old driver disks or is their format entirely new?  
My biggest gripe about the old way in RH7.x was that it's extremely 
difficult to keep track of each of the PCI ID files in each of the images.  
It's really easy to accidentally drop support for a particular card 
because the pci id gets removed from the wrong file.

That's it for now.  I'll keep on digging and see what else I come up with.

--Patrick.








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