On Fri Jan 09 2015 at 11:08:30 AM Manuel Klimek <[email protected]> wrote:
> On Fri Jan 09 2015 at 11:05:01 AM Johann Klähn <[email protected]> wrote: > >> Nevertheless the package on MELPA should be backwards compatible to >> in-the-wild clang versions, so it would have to contain some form of >> workaround for 3.4, or what do you think? >> > > So, as of r225516 we now support outputting a <cursor> element if the > cursor was given on the command line. > > What I don't understand yet is: > Why doesn't emacs correctly handle this by itself? Doesn't it update the > cursor correctly when we insert / replace code? > Figured that out myself (by checking out the xml branch of Johann's github): Emacs does the right thing unless we're within a range that is being replaced, in which case it sets the cursor to the beginning of that range (which is not what we usually want). So, I think the xml-based emacs integration works great if we use the cursor that clang-format hands back, and if clang-format doesn't hand it back when we give it -cursor= (old version), emacs' default behavior isn't too bad... > > > >> On Jan 9, 2015 10:49 AM, "Manuel Klimek" <[email protected]> wrote: >> >>> I'll update the XML format to contain the cursor position... >>> >>> On Fri Jan 09 2015 at 10:38:13 AM Daniel Jasper <[email protected]> >>> wrote: >>> >>>> The cursor being past the end of the file should not lead to an error!? >>>> Does this happen if the cursor is a the very end of the file? Should we >>>> special-case that? Alternatively, it might be easier to uses clang-format's >>>> -lines parameter instead of -offset/-length. >>>> >>>> On Fri, Jan 9, 2015 at 9:44 AM, Manuel Klimek <[email protected]> >>>> wrote: >>>> >>>>> Note that the version deletes all content of a file in case of an >>>>> error (for example, when the cursor is past the end of the file), so I >>>>> wouldn't point melpa at it yet. >>>>> I'm working on an improved version. >>>>> >>>>> On Thu Jan 08 2015 at 4:32:22 PM Manuel Klimek <[email protected]> >>>>> wrote: >>>>> >>>>>> +cfe-commits (please do not cut that out during code review, the list >>>>>> is the source of truth for code reviews) >>>>>> >>>>>> Landed as r225447. Note that I'm planning to adapt it soon to not >>>>>> replace the whole buffer, but just apply the diffs (I'll cc' you on the >>>>>> patch for review, unless you object :) >>>>>> >>>>>> Cheers & thx! >>>>>> /Manuel >>>>>> >>>>>> >>>>>> On Wed Jan 07 2015 at 6:06:08 PM Johann Klähn <[email protected]> >>>>>> wrote: >>>>>> >>>>>>> On Wed, Jan 7, 2015 at 5:14 PM, Manuel Klimek <[email protected]> >>>>>>> wrote: >>>>>>> > +If called interactively uses the region or the current buffer if >>>>>>> there >>>>>>> > +is no active region. If no style is given uses >>>>>>> `clang-format-style'." >>>>>>> > >>>>>>> > Doing full-buffer if nothing is selected is a usability problem if >>>>>>> one works >>>>>>> > in an existing codebase; if no region is selected, it should just >>>>>>> use the >>>>>>> > cursor position (which will get the current statement reflowed). >>>>>>> > The idea is that the user can do clang-format-buffer if they >>>>>>> insist. >>>>>>> Excellent point. >>>>>>> >>>>>>> > +(put 'clang-format-executable 'risky-local-variable t) >>>>>>> > >>>>>>> > Any reason not to use :risky in the defcustom? >>>>>>> Oh, I thought I already did so. I must have accidently reverted that >>>>>>> change. >>>>>>> >>>>>>> Thanks for your feedback, find attached an updated patch. >>>>>>> >>>>>> >>>>> _______________________________________________ >>>>> cfe-commits mailing list >>>>> [email protected] >>>>> http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits >>>>> >>>>>
_______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
