speed of initial downloads

[ratmice]

When running lumpy for the first time, tectonic has to download all the necessary latex packages used.
But because parsing is turing complete and tectonic downloads packages while parsing,
None of this downloading is parallelizable, it must wait for the download to complete before it can parse the next character, which then may download more stuff, etc...

the ability to do arbitrary latex inside markdown code fences complicates it further.

Perhaps there is some way to seed the cache and download the things lumpy always depends upon, (packages above/beyond what lumpy uses specified in markdown code fences, or just distribute the entire tectonic bundle 😭

It isn't even downloading much, a lot of small files, sequentially in a manner that isn't benefitting much from connection reuse since parsing often happens between.

If we can't fix the download mechanism techtonic engine itself, perhaps the bundle or cache seeding can make it acceptible... As it is it gives a fairly bad first impression.

Here is a list of ideas:

• Actually print what is being downloaded.
• Seed the cache, or download the TeX bundle on install.
• Provide some sort of pre-processor for the bundle which makes a map/graph of dependent packages, download that, and everthin listed there in