Date:	Sat, 3 Oct 2015 18:22:00 +0200
From:	Richard Weinberger <richard@....at>
To:	Alexander Kapshuk <alexander.kapshuk@...il.com>,
	linux-kernel <linux-kernel@...r.kernel.org>
Cc:	Greg KH <gregkh@...uxfoundation.org>
Subject: Re: [PATCH 3/24] ver_linux: gcc.patch

Am 03.10.2015 um 18:07 schrieb Alexander Kapshuk:
> The main objective I endeavoured to attain was to come up with an
> algorithm that would possibly result in a uniform output that would
> work across as many distros as possbile. The current implementation
> seems to struggle with that.

What that? The output if ver_linux is designed for humans.
It is not JSON, ASN.1 or XML.

> In my experience, 'sed' enables handling situations where the data
> being looked for is located in varying places more gracefully.
> 
> For example, 'gcc -dumpversion', outputs its version in a
> dot-separated numerical format. Thankfully, this format seems to be
> uniform across all the distros I have been able to test it on. So in
> this particular case, the original implementation works as expected.
> However, should 'gcc -dumpversion' change its output in the future,
> with some distros further modifying this output, so that the version
> ends up in different fields, the original awk implementation would no
> longer work. Of course, perhaps a more complex awk script could be
> written to handle this. It just that with 'sed', in my view, it would
> be a matter of adding/modifying the current patterns in a way that
> would be accommodating to the changed format.
> 
> E.g.
> 'ld -v 2>&1' output on
> Debian is:
> GNU ld (GNU Binutils for Debian) 2.20.1-system.20100303
> 
> Oracle Linux:
> GNU ld version 2.23.52.0.1-30.el7_1.2 20130226
> 
> Gentoo:
> GNU ld (Gentoo 2.25.1 p1.1) 2.25.1

I don't see a problem with these three different outputs.
Everyone can read it.

> The binutils patch:
> ld -v 2>&1 |
> sed '
>     /[0-9]$/!d    # applies to all 3 cases;
>     s/-.*//          # applies to Debian/Oracle, but doesn't affect Gentoo;
>     s/.*[ \t]//      # applies to all 3 cases;
>     s/^/binutils\t\t/
> '

Seems horrible over engineered to me.

> So far, I have not been able to come up with an awk solution that
> would work equally well across all three distros. My best take so far
> has been:
> Debian:
> ld -v 2>&1 | awk -F'[ \t\-]+' '{print $(NF-1)}'
> 2.20.1
> 
> Oracle Linux:
> ld -v 2>&1 | awk -F'[ \t\-]+' '{print $(NF-2)}'
> 2.23.52.0.1
> 
> Gentoo:
> ld -v 2>&1 | awk '{print $NF}'
> 2.25.1
> 
> Which is far from being a uniform implementation.

Why do you focus on that so much?

> I hope I am making sense here.

I fear your patch is a solution for a non-existing problem.

> I have found the proposed implementation to work well in all the
> distros I have had access to, so I thought I would share it with the
> community. If the folk here have some suggestions to make, I am
> willing to do my best to work in with them.
> 
> At the same time, I do understand that 'ver_linux' is not a tool that
> is crucial to kernel development. On this token, I do not expect, nor
> insist on the proposed implementation to be accepted. I leave it to
> the discretion of the maintainers whether or not to accept any of the
> patches.

That said, the decision is not up to me.
As I said, let's try to keep things simple unless we really need to
complicate them.

Thanks,
//richard
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists