-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

s.ross wrote:
| On Apr 19, 2008, at 11:33 AM, Phillip Gawlowski wrote:
|> Depends on your definition of large. Twitter runs on Rails. But nothing
|> like Amazon, or let alone Google.
|>
|> I have the feeling that most Rails applications are deployed in
|> intranets, to fill a particular, well-defined need, but nothing as
|> "general purpose" and exposed as Amazon yet.
|
| The implication of this post, intended or not, is that there are tons of
| large-scale public facing sites, none of them running any Ruby code.

No, it isn't. I was talking specifically about Ruby and Rails, and on
the front end of a web application, *and* adding the caveat that it
depends on what is considered "large".

Nothing more, nothing less.

(And let's face it, Rails is where you got to look if you want Ruby in
any scale of use that matters more than well within a statistical error
margin.)

Anything else you draw out of it is a hasty generalization on your part.

| There are relatively few large-scale public facing sites, period, as
| compared to the number of Web sites out there now. There's no indicator
| of how much Ruby code is in use performing non-Internet related tasks.

There are a few RQueue installations performing well in RQueue's niche:
painless setup of simple clusters, where more traditional approaches
like Beowulf are too heavyweight to use or administrate.

Does that equal large scale deployment, especially compared to the
amount of code used by Google, of which a small percentage of code is
actually Ruby? It doesn't.

| The point is, when Amazon, Ebay, and Google got their start, Ruby would
| not have been a language that came to mind as a first choice. Consider
| that these three date back to the mid-90s!

Your point? There still is nothing even *close* to Amazon, much less
Google, despite Ruby being 15 years old, nor is there anything
approaching those usage numbers.

A payment processing system using Ruport and Ruby to generate a report
for Chrysler's management wouldn't be large-scale deployment, either.
After all, at most 87 000 people are affected directly on indirectly by
that use.

No internal application alone matters. The scale is way too small.

| The corollary implication, intended or not, is that none of these sites
| could benefit from Ruby or from Rails.

Fallacious conclusion.

| The answer to that is not clear.
| Much of the code on larger scale sites has been C/C++ or Perl up to this
| point. Taking Moore's Law into account, it seems feasible that at some
| point, improvement in Ruby's performance characteristics, along with
| increase in affordable hardware capability would make Ruby just as
| obvious a choice as C/C++ or Perl were when the initial decisions were
| made to use them on these large sites. That point could be now. Amazon
| is using Rails for some of their new stuff -- not sure exactly what --
| and I know it's on everyone's radar.

Being on everyone's radar does not equal actual use. It took enterprises
almost a decade to adopt Java. And *that* had Sun's dollars behind it,
as well as the commitment of a large corporation.

Something as risky as Ruby (development could cease today, with no
further work) is fighting an uphill battle in corporate environments.
Just ask the Red Hat or Novell guys how they are feeling about that matter.

While eventually Ruby will be making inroads, it probably won't be an
epiphany at Google compelling them to throw away their existing
codebase. It'll take a new player reaching wide adoption by users, as
well as founders and funders buying into Ruby (and not just Rails).

| There are a number of Rails apps that are handling large traffic
| volumes, and Twitter is not the only one. Distilling all Ruby-backed
| sites to Twitter isn't fair to the technology, as there are millions of
| pages served a day by Rails apps, as well as by some of the less
| mainstream frameworks like merb, iowa, ramaze, etc. I don't have a
| handle on that, but it's worth noting that the absence of a huge catalog
| of "humongous site success stories" implies narrow adoption or failure.

Touch luck. Twitter is the most visible Rails and Ruby application to
date. It has buzz, hype, users, and mindshare beyond merb's or Wave's
developers. Additionally, Rails has the most visibility outside of the
Ruby community as "ruby's killer app". Compare the Rails question ending
up in ruby-talk to the merb questions, for example. The amount of
false-positives in that area are strongly in favor of Rails.

| (BTW: A number of the US political candidates, including at least one of
| the presidential ones are running Rails applications. They get lots of
| traffic :)

So? Doesn't make them large deployments, nor something that lasts.
McObamaton 2008 will disappear sooner or later. Something like Amazon or
Twtter sticks around. A blimp of usage in a year does not a trend make,
nor does it mean large scale adoption.

Notice further, that the question wasn't really about the amount of
traffic, but usage.

And my assertion regarding Ruby and Rails deployment still holds: It's
mostly intranet, for specific purposes. Nothing as general as Amazon,
Google, or other applications.

Heck, Silverlight is having larger deployments (Aston-Martin for the DBS
site, Hard Rock Cafe's Memorabilia website, Halo official community
site) than Rails or Ruby together.

All your assumptions and conclusions are based on the mix up, that
mission-critical equals large deployment. Which is humbug. A few lines
of code can be more mission critical than the whole code base together
(see Ariane V maiden explosion, STS-1 aborted first launch).

- --
Phillip Gawlowski
Twitter: twitter.com/cynicalryan

~   I think we dream so we don't have to be apart so long. If we're in
each other's dreams, we can play together all night.	  -- Calvin
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkgKYBwACgkQbtAgaoJTgL+cpwCfUTON3rQzBcqHz8/COyRtKioG
BfwAnRZb8rkJU4zulfWK4UIOBoLtax9Y
=ElfV
-----END PGP SIGNATURE-----