I have no idea how many Belgians end up in Tazzy each year, but I don't expect it to be very many.
But that's also part of the attraction : getting to know other people then one's own coutrymen. (Benidorm is definitely not for me, it's like going to SeaBruges, except the weather is better)
I don't really care where I end up : as long as English or French is commonly spoken and the hygiene is up to a certain point (meaning they actually know toilets ) I'm golden.

Well, Tassy isn't the place for you I guess. Unless you like a country with really boring landscapes , overcrowded and polluted beaches , dribbly little streams , pathetic plant life , and ugly skittish animals that you can't get close to. Not to mention the urban sprawls full of tourists.

And of course, one can't get away from the drunks...

pca

Hmmm, I see what you mean...

Dang! Guess it's off to Lloret de Mar again!

(btw : Tazzy is now most definitely in my Top3 of "places I want to see in my life" Thanks very much for the beautiful pics!)

I've got hundreds of them I took about 2GB of 2K x 1.5K pics while I was there.

pca

And of course, one can't get away from the drunks...

Yeah, they're all over here, too...

How bout a program that would build a .upgrade file from the official SB .upgrade. It could also allow you to select what apps to install on your empeg.

This sounds like a great idea. Unfortunately, the vast majority of the software distributed in the .upgrade file is in the form of an ext2 filesystem image. In order to make a new .upgrade file that contained additional or different software, you'd have to be able to create populated ext2 filesystem images. While this is trivial on most Linux machines (through the use of the loop device), it would be nigh impossible on Windows machines, which would be the most significant part of the target audience for such a monster.

Oh, come on, it's just ones and zeroes, how hard could it be?

"Oh, come on, it's just ones and zeroes, how hard could it be?"

It's the order of them that's the tricky bit...

-Zeke

Explore2fs is quite a neat program if you want to quickly access a ext2 partition from Windows. Write support isn't guaranteed to work so it may hose your partition though

- Trevor

Well for those of us with Linux boxen, how far are we from being able to do this? Obviously the program for selecting/customizing user apps would have to be written, but are most of the other ingredients there? Anyone feel like tackling this?

For those with Linux boxes, it should be fairly trivial.

I'm working on getting RAID1 working and it looks like the only way to make it work will be to modify the initrd filesystem, which means I'll have to make a new .upgrade, as I'm not aware of any other facility for uploading it, as it resides in flash (I think).

So once I get that working, I'll let you know some specific steps. However, you should be able to use Sven's UpgradeSplitterBuilder to generate one (although I haven't gotten to the point of actually trying yet). Read the man page for losetup and the ``-o loop'' parts of the man page for mount for some more information about generating the filesystem image.

Hi,

I played around with this for a while,
it's not finished, but it's a start and I'll take a few days off,
so if anyone wants to move on, feel free ....

This is untested (I don't have any machine with a serial port ),
but it should work ... so far.

Obviously S00postinstall still needs to be written,
but you can save some time using this.

anti

ps:
I hope when I come back, somebody has made this usable

Hmmm. I wasn't aware of the e2tools you're using there. It might be possible to compile those up for Win32 systems, in which case, this might all be possible for Windows users, too. I'll give it a shot in a while.

I tried compiling e2tools up with mingw, but I couldn't make it happen. If someone else can get them going under Windows, it would be a great benefit to the empeg community.

Okay, compiled it and e2tools for Windows is now here

YMMV

- Trevor

e2cp (probably the most important utility for us) seems to corrupt the filesystem, but I'll keep trying.

It worked when I tested it. How big is the file you're trying to copy?

- Trevor

512 bytes or so. On the other hand, I am taking e2ls's word for it that it became corrupted (whereas it worked before the e2cp). It might actually work in real life, though.

Edit: It is probably relevant to point out that I'm trying to copy a file from my local Windows filesystem (the README.txt that you included, in fact) to a file containing an e2fs, as opposed to solely internal e2fs copying.

Hmm.... There is definately something funny going on here... It works perfectly when I try it. Anybody else want to have a go?

This is what I get from trying to add a file to the beta13 upgrade FS image.

> C:\e2tools>e2ls pump
> bin dev drive0 drive1 empeg etc lib
> lost+found mnt proc sbin swapfile tmp usr
> var
>
> C:\e2tools>e2cp README.txt pump:
>
> C:\e2tools>e2ls pump
> README.txt bin dev drive0 drive1 empeg etc
> lib lost+found mnt proc sbin swapfile tmp
> usr var
>
> C:\e2tools>e2tail pump:README.txt
> Use filename:/directory/file to specify which file to work on
>
> YMMV
>
> - Trevor <tman_empeg@trejan.com>

- Trevor

I think e2cp is corrupting things for me too... Here's what I did:

C:\temp\e2tools_win>e2ls hda5

bin        dev        drive0     drive1     empeg      etc        lib

lost+found mnt        proc       readme     sbin       swapfile   tmp

usr        var



C:\temp\e2tools_win>e2cp README.txt hda5:/



C:\temp\e2tools_win>e2tail hda5:/README.txt

Attempt to read block from filesystem resulted in short read

Attempt to read block from filesystem resulted in short read: hda5

.

The hda5 file seemed fine until I ran an e2cp on it, then I get the "short read" error and e2tail, e2ls, etc. all fail. Seems like the file got corrupted somehow by e2cp.

That's exactly what I get. I should have been more explicit.

Edit:
Later, I tried to mount it on a Linux machine and got the following error:

EXT2-fs error (device loop(7,0)): ext2_readdir: bad entry in directory #2: rec_len is too small for name_len - offset=24, inode=11, rec_len=20, name_len=13

I also tried copy to a subdirectory, and it turns out that it made that directory unviewable. So it seems to be corrupting the directory data, possibly amongst other data.

More to come....

It also turns out that running it through fsck after e2cp'ing to it (but before mounting) loses data and doesn't even attach anything to lost+found.

What could possibly be different about your system, tman?

Bizarre, I've followed yn0t_'s steps exactly and it works for me... I'll have to try it on another system here.

I've got no idea what is different between my install and everybody elses. Probably not a good idea to use it at the moment until I have a look at it.

- Trevor