Marc-André Lureau <marcandre.lur...@gmail.com> writes: > On Thu, Oct 5, 2017 at 6:41 AM, Markus Armbruster <arm...@redhat.com> wrote: >> Marc-André Lureau <marcandre.lur...@gmail.com> writes: >> >>> On Mon, Oct 2, 2017 at 5:25 PM, Markus Armbruster <arm...@redhat.com> wrote: >>>> The QAPI schema parser has always accepted only single-quoted strings, >>>> even though JSON strings are double-quoted. Accept double-quoted >>>> strings as well, so you can write strings containing single quotes >>>> without backslash escapes. >>>> >>>> Signed-off-by: Markus Armbruster <arm...@redhat.com> >>> >>> What's the motivation to allow both? If we were to switch from single >>> to double quote only, that would make more sense. >> >> Abandoning single quotes now would require us to touch pretty much every >> line of code in the schemas. I don't think correcting quotes is worth >> wrecking git-blame. >> > > Recent (and upcoming) changes to the schema are already quite > invasive. I think we could do it, convert all strings to double-quote, > and it would help with getting the schema closer to a valid json.
Is the recent (and upcoming) churn *that* bad? Got numbers? > Fwiw, there are tools like > https://commondatastorage.googleapis.com/chrome-infra-docs/flat/depot_tools/docs/html/git-hyper-blame.html > to skip commits in git-blame. It's also fairly easy to run git blame > before the reformatting commit. Both techniques add friction... >> Sadly, the schema language is neither JSON, nor an established extension >> of JSON, nor Python. This commit brings the schema language one step >> closer to a superset of JSON. I feel "homegrown superset" is a slightly >> less bad idea than "homegrown with large overlap". >> >> Naming the schema files .json was in bad taste. >> >>> otherwise, patch looks good >> >> Ready to upgrade to R-by now? >> >> Want me to work more of my rationale into the commit message?
