Re: [Gnu-arch-users] Re: Making microbranches popular

[Tom Lord]

    > From: Miles Bader <address@hidden>


    > Checking out any new package of Tom's is a big chore because of it --
    > maybe in Tom's development environment it's all sweet and clean (I don't
    > know how he has his local disk set up), but the number of non-obvious
    > steps required to go from `I'd like to check out Tom's new pika thing'
    > to having a compilable pika tree is way too high, especially compared
    > to non-config projects where it's just `tla get foo'.

I think the truth is somewhere in between.

Yes, it takes two steps (more if you need to register and/or mirror my
archive) to get, say, a pika tree.

But, no, I think that's about as non-simple as it needs to be,
considering everything that's going on when you check out that tree.

In general:  

First, there's a huge step from someone not using revision control at
all to using revision control.   At first blush, from the individual
programmer perspective, that step can seem pretty implausible.   "Why
bother?" is an usurprising (albeit naive) reaction.

Then, there's another huge step from using really anemic revision
control to using .... oh, say ... arch.   And I think that that's a
similar leap.   And, frustrated by it, a lot of people look at
arbitrary details of arch (say, the namespace) and jump (incorrectly)
to the conclusion that that's the source of their frustration.   But
really, the leap is to recognizing a higher level of organization of
source than you've thought about previously -- and once you get there,
arch has it's arbitrary details of course -- but arch pretty much
models that higher level.

-t