On Nov 24, 12:06am, [email protected] (Izumi Tsutsui) wrote: -- Subject: Re: CVS commit: src/sys/arch/atari/stand/installboot
| One major problem of the NetBSD Project is that we don't have any | well defined procedure to get majority. | (we don't have enough activities or a person like Linus unfortunately) Yes, and voting about everything can quickly get expensive or abused. | Persons who posted patches before commit were often blocked by bikeshed, | and persons who committed changes without review (or even tests) | always won by objecting (even just "I don't think so") all post claims. | Only core might handle technical issue, but it looks core's decisions | always ignore non-technical matters (resources, marketing, users etc). One likes to think that people working towards a common goal will usually make progress in the right direction. | Unfortunately I didn't test your code at all because I objected it | (the first rev was obviously broken anyway) and it was reverted | per your approval: | http://mail-index.netbsd.org/source-changes-d/2014/11/15/msg007338.html I could not have tested; I asked you to test. | After you read other developer's post you started to claim | to strictly follow -fstrict-aliasing and use memcpy() to achieve it but | you always ignored my request (to use u_int16_t assignments consistently), | so I still cannot see what is ok for you other than memcpy (or be16enc) | since I'm not a good C programmer and it's unclear if you are ok | to use pointer conversions via (void *). I did not ignore you, I also said that your union solution is cleaner, but again it was pointed out that it would not always work... | At first you answered it didn't work | http://mail-index.netbsd.org/source-changes-d/2014/11/13/msg007326.html | and in the next you said it was legal. | http://mail-index.netbsd.org/source-changes-d/2014/11/16/msg007356.html Obviously I cannot be right on both counts :-) But at least I am batting 50%. | Anyway, per your second mail it looks you are ok to pass a structure | pointer to a function via (void *) arg so I'll add a new function to | set u_int16_t cksum at proper offset in the boot sector using | u_int16_t assignment as current abcksum() function does. I fine with it as along as it works and the compiler does not complain. | If you still don't like it, please propose a patch which satisfies | both requests as martin said, not repeating your "memcpy is ok and | -fstrict-aliasing is more important" claim. I did not propose that one but the be16enc() solution someone else presented seems more natural (and will work on LE machines). christos
