Am Dienstag, den 22.07.2014, 11:09 +0200 schrieb Holger Schurig: > Okay, another post, with less heat. > Yes, please let us keep the heat out of this argument. > I asked you specifically if a proposed solution would be ok. You > didn't answer at all. That proposed solution would still not "work" > (it won't add a memory bank, because AFAIK in sandbox there are no > memory banks at all, it just uses the hosts memory). It might compile, > however and it might be a bit of unneeded code in the "make > ARCH=sandbox sandbox_defconfig && make all" case. > I wasn't able to give any specific advice as I admitted I did not understand the problem yet. I'm aware that there are no memory banks in sandbox, but this doesn't explain a build failure. Your commit message unfortunately didn't explain this either, that's why I asked you to elaborate. Now I actually looked up the code and I think an easier solution would be to allow CONFIG_OFTREE_MEM_GENERIC to be enabled on sandbox. Would this work for you? > The tone of your mail made me think that I actually cannot convince > you, that you don't want this. Your reference to signal-to-noise made > me think this. I got the impression that you're dismissing the concept > of static checking and of code-massaging to make that easier. > Sorry, if it seemed like I wanted to trash your contribution. This wasn't my intention at all. I'm absolutely in favor of static checking but found that clangs scan-build gives at lot of false positives on other projects and I'm not really keen on adding ifdefs just for this use-case. But as I said above there may be another solution. Regards, Lucas -- Pengutronix e.K. | Lucas Stach | Industrial Linux Solutions | http://www.pengutronix.de/ | _______________________________________________ barebox mailing list barebox@xxxxxxxxxxxxxxxxxxx http://lists.infradead.org/mailman/listinfo/barebox
