diff options
Diffstat (limited to 'source/luametatex/source/luaoptional/readme.txt')
-rw-r--r-- | source/luametatex/source/luaoptional/readme.txt | 30 |
1 files changed, 30 insertions, 0 deletions
diff --git a/source/luametatex/source/luaoptional/readme.txt b/source/luametatex/source/luaoptional/readme.txt new file mode 100644 index 000000000..31b489da9 --- /dev/null +++ b/source/luametatex/source/luaoptional/readme.txt @@ -0,0 +1,30 @@ +Nota bene, + +This is the directory where optional module support ends up. Optional modules have an interface but +are not (nor will be) part of the binary. We might ship some at the context garden (like zint and +mujs) but the large one (read: with many dependencies or written in c++) have to come from the +operating system because if you use a library that is what you want: the external black box thing. +No sources end up in the distribution either, athough we will archive some. + +There will be no user modules here, just those interfaces that we provide and maintain as part of +standard ConTeXt LMTX. What users add themselves is up to them, including (long time !) support. So, +this is the canonnical version of optional. + +We might at some point add some safeguards so that we can be sure that ConTeXt is run with the +right binary because we want to prevent side effects (of any kind) resulting from a binary being +used with the same name and different features ... just because one of the objective is to have +a long term stable binary / macro package combination. Of course, what users do on their machines +is up to them. + +It might take a while before the interfaces and way we do this is stable. Also, keep in mind that +regular users never deal with these matters directly and only use the interfaces at the TeX and +Lua end. + +PS. The socket library (and maybe cerf) are also candidates for optional although cerf needs to be +compiled for windows which is not supported out of the box and sockets are way to large. We only +do optional libs that add little to the binary, a few KB at most! I'll definitely try to stick to +this principle! + +PS. Todo: move function pointers into state structures. + +Hans |