Gigantic repositories

[Tom Plunket]

> I am tracking a quite large SourceSafe repository and use only one
> repo:
> - use .hgignore to ignore what you newer will change (it's not so
> strict, you can add directories later)

Yeah that's not so much the issue for me, because when New Person
starts, I want to allow them to "sync" simply and just get
everything.  I don't mind having everything in one repo, except for
the performance implications (i.e. Tortoise slows to a turtle's pace
<g>).

> - branching does not cost anything to you, your commit's cost will
> be determined by the amount of changes what you have made on the
> working copy

I saw the docs on using links if the underlying FS supports them.  I
know that NTFS does support them but I've never heard of anyone
actually using them; does Mercurial actually use them?

> I hope this helps.

It got me to finally write an .hgignore file, but I'm still
interested in peoples thoughts on my situation.

More background:

As a game developer, I write software to target the Playstation 3,
Xbox 360, and Nintendo Wii.  Each of those platform vendors have
tools, libraries, and documentation that they distribute to allow
people to target those devices.  On top of that, other vendors
provide software (in human- and machine-readable formats) plus
documentation which people need to build the end product.  These
other vendors would be Physics or AI or Animation solutions, for
example.

The "ideal":

New Person starts work and has a new machine out of the box.  New
Person installs TortoiseHg, chooses where the repo will go, and
types (or executes via the shell) "hg clone
\\lead-programmers-machine\everything".  New Person can now edit
source files and build the game and run it on the target hardware.
(Please note that "everything" is on the order of 10s of gigs,
especially when you add in source art assets.)

thx.

-tom!