[12:05] <cr3> tepsipakki: I seem to be getting some weird preseeding behavior and I suspect it is because my preseed url contains '?' and '&'
[12:05] <cr3> here are a couple interesting lines from my syslog:
[12:06] <cr3> Dec 21 22:52:31 frontend: --> SET //192.168.2.60/certify_server?cmd=preseed&release http://192.168.2.60/certify_server?cmd=preseed&release=feisty
[12:06] <cr3> Dec 21 22:52:31 frontend: --> SUBST //192.168.2.60/certify_server?cmd=preseed&release ID //192.168.2.60/certify_server?cmd=preseed&release
[12:31] <cr3> so, instead of using cgi params, I'll try to use the PATH_INFO variable as positional arguments
[11:07] <cjwatson> tepsipakki: yeah, joeyh did it following a quick chat on IRC
[11:07] <cjwatson> needs a d-i rebuild though
[11:12] <tepsipakki> yep
[06:22] <danbe> quite a few people here:) that's what you get when you advertise in the weekly news ;-P
[07:19] <cjwatson> I didn't, not intentionally :-) But I didn't object either ...
[07:19] <cjwatson> (I advertised to mailing lists, and UWN picked it up)
[08:35] <danbe> ya, technically
[09:32] <cjwatson> cr3: the problem isn't with ? or &, but with =; if you try to use = in a kernel command line argument like key=foo=bar, the kernel breaks that up into an environment variable called "key=foo" set to "bar", which isn't what you expect
[09:32] <cjwatson> cr3: so you have to avoid = in the preseed URL
[09:33] <cr3> cjwatson: thanks for the explanation, I had already solved the problem using the PATH_INFO environment variable
[09:33] <cr3> cjwatson: in addition to solving the problem, it makes the url shorter so I have more space for kernel arguments