Date: Mon, 12 Sep 2022 14:30:21 -0700 From: Mark Millard <marklmi@yahoo.com> To: Nuno Teixeira <eduardo@freebsd.org>, FreeBSD Mailing List <freebsd-ports@freebsd.org> Cc: "se@freebsd.org" <se@FreeBSD.org> Subject: Re: Upstream changed released versions from 1.0.2 -> 3025803779 (!) Message-ID: <ED0CBDEC-8EF1-4DDE-A8AB-BB92D190F574@yahoo.com> In-Reply-To: <E88504B4-2F84-4B86-BA35-BBF661C9D9B8@yahoo.com> References: <E88504B4-2F84-4B86-BA35-BBF661C9D9B8@yahoo.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2022-Sep-12, at 14:15, Mark Millard <marklmi@yahoo.com> wrote:
> Nuno Teixeira <eduardo_at_freebsd.org> wrote on
> Date: Mon, 12 Sep 2022 18:37:05 UTC :
>=20
>> (...)
>> And at this point I'm in doubt on what to use:
>>=20
>> DISTVERSION=3D 1.0.${GH_TAGNAME} || DISTVERSION=3D {GH_TAGNAME}
>>=20
>> :)
>=20
> As far as I can tell, neither works by itself, overall.
> This is because ${GH_TAGNAME} will not in general be
> increasing: as far as I know it is just a hash and so
> will look to be random from one update to the next.
Looks like I started off with a somewhat bad assumption:
QUOTING https://github.com/aardappel/treesheets/issues/257 :
We're now using CI run id as a release tag.. these are increasing =
numbers, so feel free to use those.
END QUOTE
So however the CI run id is generated, it does provide
the ability to compare newer vs. older (given the modern
numbering technique), at least so long as the number
of digits in the ${GH_TAGNAME} string does not increase
for a pure string/textual comparison.
If the number of digits in ${GH_TAGNAME} does increase
someday, there might be problems. I've no clue how long
it took to reach the likes of 3025803779 .
> If upstream does not make the text that identifies the
> version allow sorting versions in release-sequence-order,
> it would seem that the port must invent its own release
> sequence tracking identification text. (But I'm no
> ports infrastructure expert.)
>=20
> I wonder if there are other ports that have to deal
> with such --and if there are, what they did.
>=20
> May be I'm wrong, but I thought that the ports
> infrastructure required being able to compare strings
> to identify newer vs. older.
>=20
> Having ${GH_TAGNAME} after a textual prefix that always
> allows the comparison would seem to be okay. But the
> prefix with the property looks to be essential to me.
>=20
> What a mess, even if I got the wrong understanding of
> the ports infrastructure's requirements.
>=20
>> Nuno Teixeira <eduardo@freebsd.org> escreveu no dia segunda, =
12/09/2022
>> =C3=A0(s) 19:19:
>>=20
>>> Hi,
>>>=20
>>> I forgot to mention deskutils/treesheets port.
>>>=20
>>> I've been talking with Stefan se@ and he advised me to avoid =
PORTEPOCH
>>> swithing from 1.0.2 -> gYYMMDD and use 1.0.2 -> 1.0.${GH_TAGNAME} =
where:
>>> GH_TAGNAME=3D3025803779 (release version).
>>> This way we avoid PORTEPOCH and be prepared for future 1.1.x version =
but
>>> upstream told me that they will continue to use CI run ids as =
versions.
>=20
> Wow: version identification text that is not sortable
> into a release-sequence-order relative to past releases
> at the time.
>=20
>>> Thanks
>>>=20
>>>=20
>>>=20
>>> Kurt Jaeger <pi@freebsd.org> escreveu no dia segunda, 12/09/2022 =
=C3=A0(s)
>>> 19:08:
>>>=20
>>>> Hi!
>>>>=20
>>>>>> Recently, upstream change version naming from 1.0.2 -> 3025803779
>>>> based on CI run id as release tag.
>>>>>=20
>>>>> Which port is this for?
>>>>=20
>>>> deskutils/treesheets
>>>>=20
>=20
=3D=3D=3D
Mark Millard
marklmi at yahoo.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?ED0CBDEC-8EF1-4DDE-A8AB-BB92D190F574>