[tex-live] TeXLive-CD/DVD (Installation)
dak at gnu.org
Tue May 22 10:28:51 CEST 2007
Hans Hagen <pragma at wxs.nl> writes:
> David Kastrup wrote:
>> Seems pretty pointless if you already tested the startup times. Then
>> the only advantage would seem to be that a separate binary would put
>> people at ease. But a symbolic link with a different name would
>> likely serve a similar purpose.
> the symlink (or its equivalent on windows) is just a convenience trick
> luatex --luaonly scriptname.lua
> is doing the same
Nobody who has bet on the stupidity of people ever went broke. There
is more truth in the following than many realize:
> the only reason we came up with luatexlua (can be symlink or copy)
I'd propose phasing over to "texlua" since you have
a) no luatex available when called this way, so the name "luatexlua"
is not related to the function.
b) it is a Lua interpreter with a TeX library environment (kpathsea
and font libraries), so "texlua" is quite appropriate.
c) #!texlua sounds straight to the point and not like a stuttering
aardvark with glasses and pocket protectors. One does not feel
ashamed if colleagues peer over one's shoulders when typing it.
One does not fear "Why are you using TeX as a substitute for a
scripting language? That's sick!".
> is that it's the only way to make this 'scan the first line and run
> a program' work (#....) because when that name is used the --luaonly
> is automatic
I am aware of the _technical_ aspect. Don't discount the
> when started as such, no tex mem is allocated because that only
> happens after kpse calls are used
> when using luatex as scripting engine with access to kpse, one needs
> to start it in tex mode and load the script then; no big deal
Oops. I think it would make excellent sense to still provide the
non-TeX-specific kpathsea and let the user initialize it, so that one
can implement the likes of kpsexpand in texlua without the overhead of
More information about the tex-live