On 11.04.2009 14:30, Peter Stuge wrote:
Carl-Daniel Hailfinger wrote:
clean design
Not now. Please submit your full series of patches to trac, against the 1.1 milestone.
These patches are just a refresh of a series which is almost 6 months old.
Nobody cares about flashrom 1.0 enough to actually fix the remaining design and user interface problems:
- The short command lines switches have no decided upon capitalization. - Switch combination may lead to accidental ROM content destruction (great feature!). - No way to test flashrom functionality with a single command. - Unclean generic matching handling. - We claim a generic match has unknown read, write, erase, probe results although read, write, erase do fail. I have a patch for this, but it has not been committed.
We can either say "screw users, they should read the man page every time they update" or we decide user interface questions now. And now means in the next 7 days, otherwise it will drag on longer and longer until someone is frustrated enough to simply release the current tree as 1.0.
Speaking of flashrom operation modes, only the following make sense IMHO.
- read - erase - write (without autoerase) - erase+write (only as write-after-erase) - write+verify - verify
The following don't make sense: - read+verify (it's just a reread) - erase+verify (verify needs a reference file and the erase better be checked by default)
The following are too ambiguous to be allowed: - read+erase (first read, then erase or the other way round?) - read+write (same problem)
The self-test mode has to execute: 1. read (and store on disk) 2. erase (and check if erase worked) 3. write (old contents) 4. verify (old contents)
The autoerase on write functionality (which is not present for all chips) needs to be decided upon as well.
Can we please fork an 1.0 branch and continue development on trunk?
Regards, Carl-Daniel