Re: [PATCH] Replace HTTP links with HTTPS ones: Documentation/process

From: Miguel Ojeda
Date: Sun Jun 21 2020 - 19:44:05 EST


Hi Alexander,

On Sun, Jun 21, 2020 at 4:30 PM Alexander A. Klimov
<grandmaster@xxxxxxxxxxxx> wrote:
>
> Which discussion? 93431e0607e5 ? IMAO the patches don't depend on each
> other.

The one we had the other day. It does not matter that the patches
depend on each other. It is information for whoever sees this commit.

> IMAO:
>
> * The script should not be neccessary once all of my changes[1] arrive
> in torvalds/master. Instead reviewers should say like C'mon dude, what's
> this new plain-HTTP link doing in your patch? We have 2020! Look at e.g.
> 93431e0607e5 .

In an ideal world, yes, but that won't happen unless enforced somehow.

Nevertheless, even in such a case, it would be best to have a script
to check the entire tree from time to time.

> * The program language agnostic algo description of mine should be
> enough. If it's not enough, I shall improve the description.

Then you are asking the next person to re-do the work.

> * Today I've added "If not .svg:". Imagine Torvalds merges the script,
> closes the merge window *and then* someone runs it on a random subsystem
> and discovers a missing condition. Do they have to patch the script,
> wait for the patch to arrive in torvalds/master *and then* patch the
> (other) subsystem, so they can refer to the now patched script? W/o a
> such central "rule on how to HTTPSify links" they'd just describe
> *their* algo. Or (even better) there wouldn't be much more insecure
> links, so the algo could be omitted.

I don't follow. They can patch the script if they want (or not), but
even if they do patch it, they don't need to wait for the patch to land.

> After all please show me one of the big bosses (Torvalds, K-H, ...)
> who'd tolerate to have a...
>
> * written w/o focus on maintainability
> * not documented at all
> * *Golang* file
>
> ... in the kernel tree.

It is a script, not part of the kernel building process. We already
have Perl, Python, C++, Makefile, Coccinelle...

But yes, it would be better to write it in a language that we already
have rather than add another. In particular, there is no need for a
compiled language for a script.

> If I correctly understand, you kernel devs write code so that if even
> the maintainer leaves the project, another one could just take over.
>
> How many kernel devs would read and understand (all of them I guess)
> *and maintain that Go script* of mine?

What is the problem reading and maintaining a Go script
(notwithstanding the point above)?

Cheers,
Miguel