hijack and cvs

Would it be worth having a cvs repository and cvsweb, or maybe lxr, of the hijack kernel, auto-updated?

I might do this for my own use. No promises when. Just curious if anyone else cares.

I wouldn't mind having CVS for hijack. It would help us slower and less frequent contributers. I currently find the merge process a pain to deal with.

sorry to dig this up, but i've recently been trying to compile the hijack kernel, and found it a pain in the butt to get all the patch files and apply them. cvs would be nice, or a good wget & patch bash script. hijack has a sourceforge page already, so how hard can it be?

a good wget & patch bash script.

See here.

Why not just download the complete v342 kernel tree, instead of trying to apply the 42 patches since v300 ???

And the merge issues should be near non-existent these days, now that Hijack uses differential patches and they are much less frequent than before.

-ml

yeah, i was feeling kind of dumb because i was working with v300 + 40 patches that i d/led a while back. didnt bother to look that you made a new complete base with 342. thanks.

Repository: happening now.
cvsweb: not yet

I'll send the repository information when the script finishes populating it (I started with v1 and have a script applying patches, committing with the log message which describes the patch from Mark's page, and then tagging e.g. v1, v3, etc)

cvsweb also working, but the repository not populated yet.

2 patches (iirc v70-v71 and v71-v72) missing, and one mislinked (link to v190-v200 should be a link to v199-v200). I'm trying to figure out how to deal.

Links now fixed. There is no v70-v71 nor v71-v72, but rather, a single combined patch: v70-v72

Thanks

There is no v70-v71 nor v71-v72, but rather, a single combined patch: v70-v72

Well, as of five minutes ago you had dangling links, so I was confused.

Of course, you fixed it, after I applied patches backwards from v200 and forward from v1 to generate the same patch myself. Oh well.

Edit: "Oh well" is had I guessed, the script I wrote would have dealt, and did when I reran it. Oops.

Ok, I now have a script which tries much harder to do the right thing that I'm building the repository from, and I also fixed all the patch files which had .orig or .rej files in them not to in my cached copy, so they won't spring into existance and then disappear again.

I did note my "patched to v342 from v1" had some (minor?) diffs from the v342 bzip2 tarball on the hijack website. A diff is attached.

(You can ignore the "only in" files, because I failed to run patch with -E, or whatever it is that removes files which are empty.)

After a few tweaks, hopefully the populate script has been started for the last time. It's sort of ugly, but should deal with everything needed now (cvs adding not just new files but new directories). It looks like it's taking about 2 minutes per hijack version, not shocking given that I'm tagging the entire repository for every version.

So, with any luck I should have a complete repository by the time the Pens-Kings game finishes tonight.

MMm.. That's just my "buildit" script, which I didn't realize had made it into the patches.. Safe to ignore it.

Cheers