Re: [libvirt] [PATCH 05/13] Convert (nearly) all files in src/util/ to use virReportError()

On 07/18/2012 05:52 AM, Daniel P. Berrange wrote: s/parametes/parameters/ I'm only doing a "positive review" (did the changes you make look right?), and skipping the "completeness review" (did you miss an opportunity for even more changes that should be made at the same time?). That means I'm probably not spotting the custom macros that you didn't convert, but that shouldn't stop you from applying this. As long as you are touching this, s/hre/here/ Generic question - now that we are touching 90% of the existing error messages, should we establish a convention on whether all messages should start with a lower case letter? (That would at least make us consistent with GNU Coding Standards, and while we are not a GNU project, cross-project consistency does have an advantage. Personally, I've noticed our inconsistent use of capitals for more than 2 years now, but it hasn't been high enough on my "itch radar" to scratch it.) Obviously, a syntax check would be needed IF we make a decision to standardize. I think the 'fixme' comment is no longer relevant, and should be nuked. Code like this means we will probably see a second round of patches in the future for scrubbing stupid uses of VIR_ERR_INTERNAL_ERROR into better messages for user-visible errors :) Needs a "%s" argument to silence compiler warnings when i18n is disabled. Is it worth scrubbing for error messages that end with '.'? That's another thing that GNU Coding Standards discourage. ACK. Either squash in the nits I pointed out, or we can do that as a separate patch to keep this one mechanical. As for consensus on whether to standardize error message conventions, does anyone else have an opinion? -- Eric Blake eblake(a)redhat.com +1-919-301-3266 Libvirt virtualization library http://libvirt.org