[Box Backup-dev] Reviewing code

[Charles Lecklider]

Martin Ebourne wrote:
> In the case of box though I think it makes a lot of sense to set the
> barrier high. Backup software is always under tested (in the sense that
> users don't tend to keep a close eye on it and only notice something bad
> when it fails) and over relied upon.

I think there's more than 1 barrier, at least for me:

1) Understanding what bit does what. Probably slightly lower than a lot
of other bits of software in that the design seems sensible, and mostly
the comments are helpful.

2) Source control.

a) I want to work on the Win32 version. I've never managed to get trunk
to compile, and people's personal branches are, well, personal. There's
no "Win32-stable" branch I can see, so while I could create my own and
do some work, merging that back in would be no fun at all (as I'm sure
Chris will agree by the time his changes are done).

b) TortoiseSVN is probably the least reliable bit of software I've used
for many years. I've just abandoned SVN on another project because
adding a file has about a 50/50 chance of trashing the local repository.
This pretty much rules out me creating my own branch.

3) Compiling Box. It's obviously OK on the unix side, but if trunk is
anything to go by the Win32 side leaves a lot to be desired. My main
gripes are:

a) Cygwin. For me, needing any part of Cygwin to compile something for
Windows instantly puts it in the "damn, I suppose we have to support
bloody Windoze too; let's make it work like unix" category.

b) Not using VC2k5. Again, I think MinGW is trying to do unix-on-Windows.


I don't have a problem with (1), I continue to hope that (2a) will sort
itself out with time, and I can find some way to work around (2b).

I can't live with any of (3). It's probably a good idea if Box continues
to be compilable with MinGW, but VC2k5 is what the Windows world uses. I
noticed some discussion about the licensing of the Express version not
allowing you to release a binary, but that's hardly an excuse not to use
it for development; all it takes is someone (e.g. me) with a full
version that's willing to produce release binaries.


Having said all that, I'm aware that I've contributed very little to Box
so far, so it's a bit rich to start making all sorts of demands. But
having said that, it's a chicken/egg problem for me. If (2a) and (3)
aren't sorted out I can't start doing anything, and until I start doing
something I can't really demand they're sorted out.

To extend that metaphor further than I probably should, what I'd like
ideally is for someone to say "here's an egg, here's a chicken, off you
go". Failing that, someone saying "there will be no egg" would allow me
to cross Box off the list of things I need to find time for.

-C