Re: [PATCH] doc: make HTML manual reproducible

From: Junio C Hamano <gitster@pobox.com>
Date: Dec 3, 2020, 01:33 UTC
Message-ID: <xmqqy2ifisg5.fsf@gitster.c.googlers.com>
In-Reply-To: <X8gnItjchqX4wwmt@camp.crustytoothpaste.net>

"brian m. carlson" <sandals@crustytoothpaste.net> writes:

Show 23 quoted lines

> On 2020-12-02 at 16:07:55, Todd Zullinger wrote:
>> brian m. carlson wrote:
>> > If we want this to be effective, then yes, people will need to upgrade.
>> > But if they're happy with the old behavior on ancient systems, that
>> > shouldn't be a problem.
>> 
>> Indeed.  Is it worth mentioning this at all in INSTALL?
>> Something like:
>> 
>>   -   The minimum supported version of docbook-xsl is 1.74.
>>   +   The minimum supported version of docbook-xsl is 1.74.  For consistent
>>   +   IDs in the HTML version of the user-manual, 1.79.1 or newer is
>>   +   necessary.
>> 
>> perhaps?
>
> I don't know that that's even necessary.  Anyone doing reproducible
> builds is already aware of the required versions in order to do a
> reproducible build, and I don't think the average user is going to be
> super interested.
>
> We can if you feel strongly about it, but I don't personally see it as a
> big deal.

I tend to agree.

The tool being lenient and ignoring a parameter from the future makes things very simple, and the way the patch is structured, "stable ID" is not even a build option the builder can enable. For some version of the toolchain, the option means stable ID and some older version, it does not mean anything, and that is fine.

We ship our Makefile with "CFLAGS = -g -O2 -Wall" and we do not say things like "Such and such optimizations are only available if you use GCC newer than version X.Y"; it is fine to treat the "--param" the same way here.