| 12:49:49 | bug1 | mjn3: i dotn understand your endian patch to get_header_zip |
| 12:50:15 | bug1 | if strongarm and i386 are bothe little endian then shouldnt it work on both |
| 12:50:59 | bug1 | i think cpio might have endian problems as well as its headers arent stored as ascii |
| 12:51:01 | mjn3 | bug1: morning. i just got a reply off to you. the problem was that arm pads the end of the struct, so 2 extra bytes were being read with the header |
| 12:51:08 | bug1 | k |
| 12:52:50 | mjn3 | as i wrote you, i'm not all that happy with the patch, but i didn't have much time to spend on it. so, i basicly did a quick hack to get things working... at least on arm. i haven't tested it on big endian |
| 12:55:01 | bug1 | mjn3: it seems like a bit of a minefield if a structure is different sizes on different architectures |
| 12:57:46 | mjn3 | bug1: that's why you should never read or write a struct directly |
| 12:58:22 | bug1 | im sure i read in the glibc docs that if you put __attribute__ ((packed)); at the end of the struct and it applies to every field, buti dotn know if it works in practice |
| 13:00:06 | bug1 | hmm... i can see my unions are going to be cause portability problems becasue i read the .raw part of it and union traslates it into the formated entries |
| 13:00:52 | mjn3 | bug1: that should be true, but remember that thats for the members. it was the struct itself that was being padded so that it's size was a multiple of 4 |
| 13:01:40 | bug1 | ahh yea ok, isnt there a compiler option to control that ? |
| 13:02:57 | mjn3 | no idea... and even if there is, i try to avoid such things since it would be nonportable by definition. some of the uCilbc stuff i've done also builds with bcc, the 16-bit compiler used for ELKS |
| 13:05:08 | bug1 | how about the aproach cpio uses, it reads a 110 byte block and uses sscanf to convert it to the seperate fields.. that should be portable shouldnt it ? |
| 13:18:55 | mjn3 | bug1: not sure. i'd have to look at it and i need to be somewhere in 30 min. don't have time to look at it right now |
| 13:28:39 | bug1 | mjn3: fair enough, i have to cleep now anyway... talk to you later |
| 13:28:43 | bug1 | sleep |
| 16:04:43 | mjn3 | BZFlag: morning. i sent an email about usbdnet to the address you gave me. no reply as of yet |
| 16:25:29 | BZFlag | cool. copy me on any others you send so I can see how stuart reacts. ;-) |
| 16:25:59 | mjn3 | ok |
| 23:17:31 | mjn3 | bug1: glenn, i must be doing something wrong. but the few times i've tried, i haven't gotten cpio in unstable busybox to work |
| 23:22:04 | bug1 | mjn3: hmm... on i386 |
| 23:22:17 | bug1 | mjn3: cpio -iF filename.cpio |
| 23:22:24 | bug1 | that extracts |
| 23:22:54 | mjn3 | i'll try it again |
| 23:23:06 | bug1 | i havent tried the unstable branch for a while, the version in my new unarchive code works if you want to give that a try :) |
| 23:24:31 | mjn3 | ok.. later though. got to go wash dishes... |
| 23:59:46 | bug1 | mjn3: one think about cpio is that there are a few different standards |