Hacker News new | past | comments | ask | show | jobs | submit login

In this paper[1], they say that much of the latex running time is eaten up in the startup and shutdown of the program and the actual typesetting time is in the microseconds.

A long-running process (daemon) is therefore a much more efficient way to run (La)TeX.

[1] www.tug.org/TUGboat/tb27-0/fine.pdf

""" On my current 800 MHz PC, the command

  $ tex story \\end
takes about 0.137 seconds, while

  $ tex \\end
takes 0.133 seconds. The first command typesets a small page of material; the second does nothing but start TEX and then exit. Thus, typesetting the small page takes about 0.004 seconds. """



It strikes me that TeX ought to take a page from Lisp interpreters and dump core after loading all the standard configuration files, so that future runs won't need to waste so much time on initialization.


It does, that's how LaTeX starts up and why you have to run it as latex instead of tex. Ditto texinfo, and probably ConTeXt. For a variety of reasons the TeX engine still has to reload a lot of stuff because it might possibly have changed, which is a suboptimal situation; but there is a concept of a "dump" state that has most of the kernel initialized. I suspect people don't care enough to work around it, and the people most likely to do the work (TeX kernel hackers) are also the people who are most likely to be inconvenienced by it not noticing the tweaking they're doing.

Which is not to say it's not a good idea here.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: