Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 25 Jul 2015 12:22:13 -0400
From:      Allan Jude <allanjude@freebsd.org>
To:        Alexander Leidinger <netchild@freebsd.org>, Johannes Jost Meixner <johannes@meixner.or.at>
Cc:        rene@freebsd.org, dchagin@freebsd.org, freebsd-emulation@freebsd.org
Subject:   Re: [CFH] Allan's 64bits CentOS ports
Message-ID:  <55B3B7B5.5000009@freebsd.org>
In-Reply-To: <20150701152052.Horde.XdxXFlk6nCjUyNqwVIVyVQ9@webmail.leidinger.net>
References:  <20150701152052.Horde.XdxXFlk6nCjUyNqwVIVyVQ9@webmail.leidinger.net>

next in thread | previous in thread | raw e-mail | index | archive | help
This is an OpenPGP/MIME signed message (RFC 4880 and 3156)
--ktarTc57GnOcGWx39uC5IVFjbvnVT8FmE
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable

On 2015-07-01 09:20, Alexander Leidinger wrote:
> Quoting Johannes Jost Meixner <johannes@meixner.or.at> (from Wed, 01 Ju=
l
> 2015 07:53:25 +0300):
>=20
>> Allan could use some help reviewing his suite of CentOS 6.6 64bit port=
s.
>>
>> https://reviews.freebsd.org/D1746
>=20

Sorry for being late replying to this

> I had a quick look at comments on the top of the page and the diffs of
> bsd.linux-apps.mk and bsd.port.mk.
>=20
> What I'm missing here (I may have overlooked it, it's the first time I
> have a look at reviews.freebsd.org) is a short explanation for the
> rationale of the design decissions (see my questions below).
>=20

Caveat #1: I have no idea what I am doing. I just poked stuff until it
worked. There were very few "design decisions"

> The very first questions which come to my mind are:
>  - Why is this embeded into the existing ports instead of having it as
> seperate ports?

Cloning every linux port seemed excessive. Especially considering the
only difference in the ports is swapping i686 for x86_64 in the name of
the .rpm, and the lib path changing to lib64

My main goal with combining them was easing the amount of work in
maintaining the ports and keeping them up-to-date

>  - Would seperate 64bit ports make the infrastructure less
> convoluted/complicated (KISS)? Yes, more ports, but the Mk
> infrsatructure is already at a complexity level where not much people
> are willing to touch it, and with this I fear it will be just too much.=


There isn't that much magic in the Mk stuff, other than moving a few
things that would have had to be in each separate port.

>  - Can I install 64bit and 32bit in parallel with this approach (I have=

> to admit, it depends if the 64bit linuxulator is going to a different o=
r
> the same /compat/linux directory but I haven't checked that, and it
> depends on how centos is build in this regard, so no idea if this makes=

> sense)?

Not currently. I had thought about making the 64bit port install both
the 32bit and 64bit files, since the 64bit repo contains slightly
different versions of all of the 32bit .rpms that appear to have any
conflicting files removed, so are designed to be installed along side.

>  - Is it a good idea to play around with the portname here (ok, this
> fits into the first question)? My concern here is that some ports playe=
d
> around with the port name in the past and got slowly converted to
> something without the name-mangling because we learned that it was not =
a
> good idea.

This may be true. I developed all of this in isolation, these are maybe
by 6th time working on the ports tree.

>=20
> Apart from that I have to admin that I don't like that
> OVERRIDE_LINUX_BASE_PORT is used to check for 32bit or 64bit installs o=
f
> the linux base. IMO it makes more sense to have a sort of "I want to
> have a XXbit linuxulator" variable: would be more end-user friendly and=

> better self-explaining code (related to KISS).

This was stuff that was already there, and I just added the 64bit option
to it. When I was originally developing this stuff, the default
linux_base port was not c6, so this may have been overcome by events.

>=20
>> What I'd like to see is, moving the Mk/ infrastructure to the point
>> where it can support future, upcoming architectures -- think CentOS 7,=

>> recent Fedora version (only the ones that are supported for more than
>> 6 months), etc.
>>
>> I saw a working port of CentOS7 on GitHub, and a working Port of
>> Fedora 19 somewhere... but I don't recall the links. Check
>=20
> Can someone dig out the link for the CentOS7 port? 24th to 26th there i=
s
> the DevSummit in Essen/Germany and I thought about the possibility to
> have a look at CentOS7 ports (if I don't find something less painful to=

> work on) and it doesn't make sense to re-invent the wheel.
>=20

Google finds no results for "linux_base-c7". My original start had
simply been: do a minimal install of centos7 in a vm, and extract the
list of rpms that were installed, and update a copy of the linux_base-c6
port with those package versions etc.

Note: CentOS7 is 64bit only (solving the "how do we deal with 32bit and
64bit at the same time" problem). However, CentOS7 also uses systemd,
and may rely on other newer kernel features our linux_compat may not have=
=2E

> Bye,
> Alexander.


--=20
Allan Jude


--ktarTc57GnOcGWx39uC5IVFjbvnVT8FmE
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (MingW32)
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=KASU
-----END PGP SIGNATURE-----

--ktarTc57GnOcGWx39uC5IVFjbvnVT8FmE--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?55B3B7B5.5000009>