Hello,
I read recently Bowkett's "Why I Program In Ruby (And Maybe Why You
Shouldn't)" (
http://gilesbowkett.blogspot.com/2007/11/why-i-program-in-ruby-and-maybe-why-you.html),
and I have given it a little bit thought. Ruby is indeed what Matz has aimed
for - a blur between art and craft. Guys like why make my day with their
posts and I watch with awe some of the projects that are using Ruby. It as
language makes me feel unbelievably comfortable and makes programming even
fun (I am Msc (econ), not an engineer!) but still something is wrong.
It is not just about language after all. The language is great already, but
it is not enough. The standard library is the problem. Ruby is nice for
small demoing, replacing shell scripts and such but when you try it for
something more complex than what that nice "try Ruby in your browser"
application was meant for you get into serious trouble. It is something like
trying to barbecue some chunky bacon using a toothbrush and a dead squirrel
as heat source. You get instantly dragged into installing extra native
libraries and their bindings (process that means unreliability and extra
work for packaging etc if you wanted to redistribute it) and attempting to
understand extremely poorly documented classes. Ruby is a lot like one of
those screwdrivers with detachable tips. The handle is awesome but there are
no tips for it.
Let's take the standard library under scrutiny. First of all, Ruby lacks a
GUI toolkit. For historical reasons it comes with the "cross-platform" TK.
That's nice if the cross-platformity meant the same as "looks and feels out
of place and plain amateurish on every platform". After seeing something
like Shoes the TK really makes baby jesus cry. Ruby is entirely unsuitable
for GUI application development. No one really wants to install all those
3rd party hackish dependencies and pray constantly that all your customers
etc manage to keep them runnin as well. Too much complexity and risks.
Drb is awesome, but try the crypto (SSL) version using proper keys and
certificates (which you require for nearly every REAL use case anyways).
Have fun hunting some sane examples and documentation. The same goes for the
other networked classes as well. Their even slightly more advanced use cases
are entirely undocumented or as with some plain impossible. Try implementing
SSL+xmlrpc server and client and come back in 6 months or so when you have
succeeded. Very basic and expected features, all missing. All around.
Although Ruby knows how to build documentation, where on earth is my search
function? Frames, butt-ugliness and awful usability is all that I can see.
You could have expected that sort of output in the 90s perhaps but not
nowadays, not in a modern environment! On top of that most of the std-lib
documentation is written for people who already have used those libraries.
API documentation is just a tiny part of what should be provided. The lack
of examples, rationaile and verbosity are lethal for attempting to
understand that jungle of hastily ported and imported mess. For example you
click on something like tracer and the most describing thing there is
"tracer main class". Uhh gee. Thanks. Very useful. I think tracer is a
bullet with incendiary chemicals added for burning while flying towards
target.. Still. The whole pile of "documentation" is like that. Unfriendly
and plain useless.
There's a lot of legacy in Ruby. For instance no real utf-8 (come on, UTF-8
is earlier 90s invention already!), stuff like webrick (there are better
alternatives nowadays), and so on. In fact I think there is nearly complete
lack of development outside the very core of the language. In fact I think
it was never developed far enough before it really stopped again. There is
no sense of it ever having any real direction and determinant force,
especially on what comes to the standard libraries which are mostly just
plain childish and protected by insane amount of change resistance and lack
of resources or will.
All in all, Ruby could really shine. But I think it will never do that. It
is a real shame because I kind of like it. Also I fear 99% of the people
that bothered reading past the title will not get what I tried to say. Do
not comment on any examples or individual problems. If you do, you have lost
the point entirely. The symptoms should not be fixed. The root causes
should. The root cause is: complete lack of vision, leadership and will to
make Ruby a complete and modern environment for most common tasks. At this
moment it is more of an academic and assembly-required thing that can not
(not defined by "ability" but by sanity of using it as a tool for such task)
be used for nearly any non-childishly simple project.
(Rails might be an exception though, but the web frameworks seem mostly the
only sane domain for Ruby at this moment.)
