Subject: Re: [RFC] mk/latex.mk to handle latex dependencies
To: None <email@example.com>
From: Antoine Reilles <tonio@NetBSD.org>
Date: 11/10/2005 13:49:18
On Nov 10, 2005, at 6:15 AM, Min Sik Kim wrote:
> On 11/3/05, Antoine Reilles <Antoine.Reilles@loria.fr> wrote:
>> The file now is mk/latex.buildlink3.mk
>> I removed the FIRST_ACCEPTED logic, so now if a package do not
>> accept the
>> default version, which is defined by the user settable
>> LATEX_VERSION (defautls
>> to "teTeX2"), it will fail, and require the user to explicitly set
>> LATEX_DEFAULT to something the package can accept.
> Thank you for working on this. I have a couple of comments.
> Shouldn't these LATEX_* variables be TEX_*? Selecting teTeX version
> using LATEX_* through tex.buildlink3.mk doesn't look consistent.
Why not. You're right, using TEX_ looks more consistent.
> Also, I think it would be better to split the version selection part
> from buildlinking (e.g., print/teTeX/version.mk) since very few of
> tex-* packages need buildlinking. See ruby and python for example.
I'm planning to convert print/teTeX in a special package as for
python. At this stage, it may be good to separate those concerns. For
now, i have chosen do to the buildlink in all cases, because it
involves only very few files. Maybe print/teTeX could contain thing like
tex-application.mk, for applications using tex at run time, or
compile time, and
tex-modules.mk, for tex extentions.
That one could also take in charge running mktexlrs, and also helpers
for modifying the texmf.cnf file