Re: [PATCH gwproxy v8 2/2] gwproxy: refactor code base to add experimental raw DNS backend

[reyuki <reyuki@gnuweeb.org>]

On Sat, Sep 6, 2025 at 1:47 PM Alviro Iskandar Setiawan wrote:
> On Sat, Sep 6, 2025 at 11:33 AM Ahmad Gani wrote:
> > I thought it was necessary to mark that a certain block belongs to the
> > experimental raw DNS feature. If I reduce the use of these marks, it might
> > introduce unnecessary code into the compiled version when this experimental
> > feature is disabled.
>
> It's necessary, but you're overusing it. Your version clutters the
> codebase so much. You had better have empty local functions which are
> compiled to nothing instead of #ifdef/#endif everywhere. Look at
> https://git.kernel.org/pub/scm/linux/kernel/git/axboe/liburing.git/tree/src/include/liburing/sanitize.h?id=5baeec1e1200b88da1d84ba978605728d791d693

I have reduced it to 17 using a similar way with the liburing way,
is it still too much? you can take a look at this branch [1]

FYI, I've pushed some commits implementing your proposal and Sir Ammar's,
except for the third point in Sir Ammar's proposal. As Sir Ammar noted,
DNS queries are short-lived objects. If the concurrent queries exceed
2^16, I think we can assume we're being DDoS'ed and can safely ignore
further requests.

If some of you think exceeding 2^16 is realistic in production and are
willing to reproduce such a scenario and show me results where it
actually happens, I'll reconsider implementing the third point. For now,
I believe in the YAGNI principle.

[1]:
https://github.com/realyukii/gwproxy/tree/archived-experimental-raw-dns