# Local texmf tree vs. TEXINPUTS

Integrating Local Additions provides a nice explanation of how to manipulate TEXINPUTS or register a local texmf tree.

Q: When integrating local additions, is there a reason to prefer one approach over the other?

I'm guessing that TEXINPUTS is just regarded as the quick and dirty approach while registering a local texmf tree may be a better idea if you have a lot of files within a logical directory structure to which latex needs to be pointed.

If I'm missing something here I'd love to hear about it :)

This question is a follow-up to my post Texmf tree with MikTeX

-

TeX itself uses TEXINPUTS and it can cause an error if it is also set by the user. An almost secure way is to use a local texmf tree. With MiKTeX you can have several of them. For Linux you have TEXMF, TEXMFLOCAL and TEXMFHOME. The latter do not need a texhash´ command, it is searched recursively by TeX.
I have never heard of using TEXINPUTS causing trouble if used correctly. It is intended to be used by the user, surely? (And by used correctly, I mean typically with a trailing colon, to insert the directories spesified by the texmf.conf file(s) at the end.) –  Harald Hanche-Olsen Mar 1 '11 at 12:04
I know that on MikTeX 2.8 TEXINPUTS` is not even initially defined. The user must define it and initialize it to the path of interest. I played with it---adding, deleting, etc. No issues so far. –  lowndrul Mar 1 '11 at 15:09