Policy always was and has been, that we try to achieve consistent guidelines for new files and in general the guidelines for each file should be reflecting that files style.
Although I do appreciate applying consistent styles, I acknowledge the fact that we have a really old code base in some places and we shouldn’t force a change everywhere because of it. Can also see, that that would be neat with some opposition. So in general, although I appreciate having new files apply to style guides, I would keep the existing ones as is Cheers, Roland > Am 01.03.2022 um 19:23 schrieb João Valverde <j...@v6e.pt>: > > On 01/03/22 17:45, David Perry wrote: >> Hi all, >> >> Bottom line up front: how much do people care about the formatting of >> Wireshark's source code? > I would like to have indentation harmonized (and enforced consistently) > across the entire C code base. Preferably 4-space. > > Don't care so much about other style issues. I don't think that needs to be > enforced. > >> Background: I'm looking into [#17253][1]. It's chiefly about removing editor >> modelines from the footer of each source file in favour of just using >> `.editorconfig` files. But by extension it's also about removing the >> exceptions from `.editorconfig` files and making the formatting rules >> consistent across files. >> >> I took a manual pass at harmonizing the formatting of the C files in the >> root of the repo and that was painful, so I researched automatic approaches >> for the rest of our code. [Clang-Format][2] seems to be a popular approach >> for this sort of thing. >> >> Automatic code formatters in general, and clang-format in particular, are >> rigid and somewhat naïve in how they do things. This is in contrast to the >> flexible formatting practices we use. That's not a huge deal if we just want >> to reformat once to harmonize our indentation levels and whatnot, and then >> return to manually formatting based on the new standard. >> >> On the other hand, a comment on !6298 suggested that automatic reformatting >> could be integrated as a pre-commit hook and/or a CI step. That... also >> isn't a huge deal, I guess. We'd have consistency across files at the price >> of slightly less formatting freedom. (And of having another developer >> prerequisite to install, if we did it as a pre-commit hook.) >> >> But it's a decision that should be made by the dev community as a whole. So >> what do you folks think? Is consistent formatting important to you? Would >> you like to see it enforced with an automatic formatter? >> >> (My proposed `.clang-format` file is in [!6298][3] and aims to capture the >> most common practices used across the codebase. Please use that MR for >> discussions about specific formatting details. This email is for the general >> discussion of whether/how to apply and enforce formatting.) >> >> Thanks for your time, >> >> David Perry >> he/him >> >> [1]: https://gitlab.com/wireshark/wireshark/-/issues/17253 >> [2]: https://releases.llvm.org/13.0.1/tools/clang/docs/ClangFormat.html >> [3]: https://gitlab.com/wireshark/wireshark/-/merge_requests/6298 >> ___________________________________________________________________________ >> Sent via: Wireshark-dev mailing list <wireshark-dev@wireshark.org> >> Archives: https://www.wireshark.org/lists/wireshark-dev >> Unsubscribe: https://www.wireshark.org/mailman/options/wireshark-dev >> mailto:wireshark-dev-requ...@wireshark.org?subject=unsubscribe > > ___________________________________________________________________________ > Sent via: Wireshark-dev mailing list <wireshark-dev@wireshark.org> > Archives: https://www.wireshark.org/lists/wireshark-dev > Unsubscribe: https://www.wireshark.org/mailman/options/wireshark-dev > mailto:wireshark-dev-requ...@wireshark.org?subject=unsubscribe ___________________________________________________________________________ Sent via: Wireshark-dev mailing list <wireshark-dev@wireshark.org> Archives: https://www.wireshark.org/lists/wireshark-dev Unsubscribe: https://www.wireshark.org/mailman/options/wireshark-dev mailto:wireshark-dev-requ...@wireshark.org?subject=unsubscribe
