Hi Sascha On Wed, 2014-05-21 at 15:51 +0200, Sascha Hauer wrote: > On Tue, May 20, 2014 at 07:27:55PM +0200, Christoph Fritz wrote: > > Command 'findstr' can be for example used to find the string > > "MAC=1C:BA:8C:F3:82:BB" in file /dev/eeprom0 to set the > > appropriate variable: > > > > $ findstr -o 4 -l 17 -t eth0.ethaddr MAC /dev/eeprom0 > > > > Usage: findstr [OPTIONS] <STRING> <FILE> > > Find string in file and print it > > -o <offset> set offset of string which gets printed > > -l <length> set length of string which gets printed > > -t <var> print into variable instead of stdio > > I wonder how useful such a command is. I mean normally I would > expect a MAC address to be stored in binary format and not ASCII. Yes, and if a user wants to set it to another value, there is variable 'ethaddr' in 'env/network/eth0' which can be changed. The use case here I'm facing: User wants to change 'ethaddr' and store it in eeprom instead. If the whole /dev/eeprom0 would be a barebox-script, it could be processed, but I can't rely on. > Matching 'MAC' (what about the '='?) seems not very robust. >From the example above, the offset is 4 and a '=' or any other char gets bypassed. > What about strings with dynamic lengths? A more robust and flexible approach I have in mind is this: Find string xy and delimiter z in file and optionally execute what is in between as a command. > If the command is primarily intended for the MAC address then maybe > a more dedicated command or a piece of code called from board code > is more robust. That's what I'm already doing here, but want to get a more general solution for all kinds of settings. > If the command is intended as a general > eeprom-variable-store then I think it's not flexible enough. What do you think about the more robust approach from above? Thanks -- Christoph _______________________________________________ barebox mailing list barebox@xxxxxxxxxxxxxxxxxxx http://lists.infradead.org/mailman/listinfo/barebox
