|
Boost-Build : |
From: Robert Ramey (ramey_at_[hidden])
Date: 2007-11-01 14:12:02
>> This is really confusing and annoying behavior.
>
> This is by-design,
I'm aware of that. In fact, that is the substance of my complaint.
Its not that bjam is not a huge accomplishment in a technical
sense or that its this or that bug or whatever. My view
is the whole goal of a "smart program which figures out
what the user really should want" is a bad idea. Take this
to extreme an one ends up with surprising behavior all over
the place. I realise its not just bjam its a lot of programs
these days. MS Word is the program which manifests
this behavior in the extreme.
Perhaps bjam should be spun off as a commercial
product. Suppose a licensee of this new technology
were to build a soda vending machine.
a) display shows a huge selection of possible sodas
including sweet yes/no, sour yes/no, caffeine yes/no,
etc.
b) I step up an just want a classic coke. But of course
there is no classic coke there. So I press a bunch
of selections which I hope will result in what I want.
c) first of all it warns me that such a combination of
selections will result in something that is unhealthy for
me and will increase my carbon footprint.
d) But, since its only a warning, it makes a selection
for me.
e) which results in some weird thing I never imagined.
Since the machine is so sophistcated it can produce
more combinations of sodas than it can produce labels
for so I get can with no label. Well, no matter, take
it on faith that its what I want, after I specified what
I wanted.
d) Damn - no bubbles - tastes awful.
e) go back to b) and modify my selections.
f) after a few iterations, I just go to the water fountain.
So, maybe the machine won - and I'm am healthier for it.
I don't mean to pick on bjam specifically. This stuff
is everywhere and it's starting to drive me crazy. But
my point remains. The problem of bjam is not its
implementation, its syntax, etc. These things could
all be improved but it won't fix the fundamental problem.
The fundamental problem is that it tries to do more than
it can really do.
> and I don't recall many complains about that.
LOL - well, here's one.
> I'm actually not sure what kind of reply you expect.
I don't really expect a reply. I'm just relieving myself
of the agravation of having to deal with this. Sorry
Robert Ramey
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk