Can't update feeds?

[chet]

On Jul 16, 2008, at 12:12 PM, Rodger D. wrote:

Why is mod_proxy working with mongrel such an exercise?
That’s it as a whole, 7 whole lines. Add that to your apache
configuration in a Virtualhost area for your blog and startup typo
and you should be golden.
At which point you wonder why everything is running so slow, and you
discover that mongrel really, really sucks at delivering image files
and the like. So your 7 line example works if you want horrible
performance with even a trivial number of users.

I absolutely agree. Mongrel does suck for delivering images, however
that is part of scaling and proper design. If you use Swiftiply it
does take some of the pain away, but Mongrel has certain problems (or
should I say rails really?) images, uploading images… bone jarring
pain. If you have let’s say a gallery, and you want it to send 80
thumbnails to a user. Well that’s just inefficient when you can have
Apache or whatever web server you have handle the images in 1/32nd of
the time. FastCGI really does not make that pain go away, it’s
usually easier to host your images on apache and have your ‘rails app’
reflect where the images are to be pulled. Lessens the amount of
requests per second to Mongrel(or whatever you use) which can make
your web app smoother as well as allows other users to take up the
requests that the images were sucking up.

I’ve tried Gentoo and it’s worked excellent also, so perhaps some
research is in order?

Actually, I’ve used Ruby on a number of the Linux problems, and the
interaction of Gems and Ruby is a problem on all of them. A snide
and condescendng tone does not change this fact, it merely convinces
people they don’t want to bother using typo.

That would be a side effect, some call it ‘maturity’, I prefer to
consider it stagnation. It’s been many years since there was a major
Perl version release. When Perl6 starts coming in Linux Distributions
you’ll feel the same exact pain as you do for Ruby. I like to call
this side effect “People telling you what to run, how to run it, and
what version to run”. I dislike that side effect because if you want
to run your own version it becomes painful (even for Systems
Administrators it’s painful) and after a certain point you have to
decide when it’s too much and you need a change so you don’t have to
keep doing this.

I disagree that it’s a typo issue really, the issues you are feeling
are more Ruby issues and Ruby on Rails, not Typo. You can ask Someone
to update this and update that and make Ruby a better experience, but
if they don’t give a damn to do that … There’s not much to do. I
consider that one of the pains of running a Binary Distribution. They
attempt to lock you into what they offer you, and make it a hassle to
go beyond that.

[chet]

Chet F. wrote:

On Jul 16, 2008, at 6:10 AM, Scott L. wrote:

I’ll certainly agree with that. Getting mongrel working with
mod_proxy was essentially an exercise in Google and reading blogs.

Why is mod_proxy working with mongrel such an exercise?

Beats me. Perhaps you should refer to the first portion of my reply to
you last night.

Explanation by analogy would be helpful; “Mogrel fills a role similar to
that of Tomcat for JSP applications” would be a good starting point for
most people with experience with web apps.

The best explanations of how to make it all hang together well I found
were at

http://blog.codahale.com/2006/06/19/time-for-a-grown-up-server-rails-mongrel-apache-capistrano-and-you/,

http://jonathan.tron.name/2006/07/26/apache-2-0-x-mongrel-mod_proxy-mod_rewrite-configuration

but one of those is talking about another Rails blog tool, of course.

This would require Typosphere to be online and updated from time to
time.

It’s definitely Ruby’s problem if PHP, Perl, Python, etc., are all
running fine out of the box.

It is a problem with Ruby & Gems specifically that Gems don’t integrate
as smoothly as extending Perl with non-packaged CPAN modules does on
major Linux distros.

[chet]

On Jul 16, 2008, at 6:43 AM, Chet F. wrote:

involved, what the alternatives to a Mongrel configuration are, and
why one might choose one approach over the other are ALL questions
that need to be addressed. I made that clear in my prior post.

We are comparing an Apple to a Pear, LAMP is not the same as LAMR or a
Ruby on Rails install. Please stop comparing it, you are doing
nothing useful by doing that.

What is mongrel? Let’s see here if this quote satisfy’s you.

"Mongrel is a small library that provides a very fast HTTP 1.1 server
for Ruby web applications. It is not particular to any framework, and
is intended to be just enough to get a web application running behind
a more complete and robust web server.

What makes Mongrel so fast is the careful use of an Ragel extension to
provide fast, accurate HTTP 1.1 protocol parsing. This makes the
server scream without too many portability issues.

See mongrel.rubyforge.org for more information. "

A bit technical, but Mongrel is a Web Server. Like Apache, or nginx
or lighttpd. Except that all it cares about and knows about is the
application in which it is running. Technically you can run it on
port 80 (if you run it as root) and remove the whole extra web server
from the equation. Mongrel is for Ruby what mod_php is for Rails (bad
analogy). It does all the dirty work and forwards the end product to
Apache and then to the end-user. You can compare it to Tomcat or
Jetty I guess, however that’s not here nor there.

this point my Typo runs (just not in the way I really want it do;

– Typo does not work properly with MarsEdit or other stand-alone
editors. This is a show-stopper for me.
I also have some other outstanding questions regarding updating my
Typo, and verifying the version I have, but those are in another
mail I posted early yesterday, I believe, and are part of a dialog
with Frederic.

• How do I verify what version of Typo I have?
• How is it best to upgrade Typo? What specific steps should be
  taken, and why?

Frédéric has pretty much gone into the other problems you addressed.

However, the best method of updating/upgrading typo depends on the
backend.

Let’s say you use SQLite3

1. backup your databases in db/ to a separate location
2. unarchive the new version of typo in a separate directory (for
   staging or testing purposes).
3. copy the backed up databases (do not move them, we want to keep a
   pristine copy still somewhere) to db/
4. verify you are using the proper environment (e.g. PRODUCTION /
   DEVELOPMENT …) and then run rake db:migrate
5. start typo on a different port then the original typo is running on.
6. Browse to the site by ip:port and verify if everything worked as
   planned, and if it did not discuss how it did not work properly, and
   if it did work how it can work easier.

… Now let’s say you use MySQL.

1. mysqldump your typo database for backup purposes.
2. mysqlhotcopy -u root -p typo typo2
3. fixup the grants if need be so your typo login can see typo2 (if
   you use the root account for typo you can skip this step)
4. unarchive the new version of typo, and setup database.yml to see
   typo2 and the proper credentials and settings.
5. rake db:migrate
6. start typo on a different port and test it out.

Of course you’ll have to reinstall your plugins and themes. I believe
as a standard practice it’s best to set the theme to default and
disable your plugins before you do the backup. I don’t know if
Frédéric has resolved the technical details of plugins missing and
themes missing and having to goto the admin interface and fix it, or
delve into the MySQL database and disable plugins that way.

But that’s a pretty short write-up.

[chet]

On Jul 16, 2008, at 5:38 PM, Scott L. wrote:

a Ruby on Rails install. Please stop comparing it, you are doing
nothing useful by doing that.

Are you really saying you can’t compare a Ruby app with a LAMP app?
That’s ridiculous. How else can someone decide between Typo and MT and
WP and etc?

No, you’re just wrong. It makes PERFECT sense to compare the
experience of setting up and using Application X with that of using
Application Y if both X and Y are competitors in the same market
(i.e., blogging software).

See mongrel.rubyforge.org for more information. "

A bit technical, but Mongrel is a Web Server. Like Apache, or nginx
or lighttpd. Except that all it cares about and knows about is the
application in which it is running. Technically you can run it on
port 80 (if you run it as root) and remove the whole extra web
server from the equation. Mongrel is for Ruby what mod_php is for
Rails (bad analogy). It does all the dirty work and forwards the
end product to Apache and then to the end-user. You can compare it
to Tomcat or Jetty I guess, however that’s not here nor there.

Why on earth would you need to run a second web server? That seems
like a really bad idea, frankly, hence my annoyance that the most
obvious question (which boils down to “WTF?”, essentially) isn’t
addressed.

3. copy the backed up databases (do not move them, we want to keep a
   pristine copy still somewhere) to db/
4. verify you are using the proper environment (e.g. PRODUCTION /
   DEVELOPMENT …) and then run rake db:migrate

What does rake do in this context?

5. start typo on a different port then the original typo is running
   on.
6. Browse to the site by ip:port and verify if everything worked as
   planned, and if it did not discuss how it did not work properly, and
   if it did work how it can work easier.

Of course you’ll have to reinstall your plugins and themes.

!!!

To put it mildly, that’s a bit bizarre and very unfriendly to the user.

I believe as a standard practice

Maybe for Typo. Not for anything else I use.

Chet F.

“When you’ve got an RV, a jet pack, and a monkey you really don’t need
much actual content” - KS

[chet]

On Jul 16, 2008, at 6:38 AM, Chet F. wrote:

Beats me. Perhaps you should refer to the first portion of my reply
to you last night.

It’s clearly a problem, though. It’s also a problem that the purpose
of Mongrel isn’t made clear; you just have to take on faith that
it’s something you need to do based on the sketchy installation guide.

After reading your posts I’m not sure if you are using Debian, or
what… however Here’s a few articles courtesy of the folks at mongrel.

http://mongrel.rubyforge.org/wiki/Debian
http://mongrel.rubyforge.org/wiki/OSX
http://mongrel.rubyforge.org/wiki/Lighttpd

(Formatting seems to be off on the last 2) however you get the idea.

It’s definitely Ruby’s problem if PHP, Perl, Python, etc., are all
running fine out of the box.

Here, you’re defaulting back to a knee-jerk defense of what is
clearly your pet language. That has no place here. Compared to LAMP-
stack stuff, RoR applications are much harder to set up and deploy.
They require a totally different approach, and that approach is very
poorly documented. This isn’t a controversial statement.

Do you have the latest version of Python 2.5? what if you want to
deploy a Django application that requires Python 2.5? and some feature
that was not included in your build of python. knee-jerk defense
about Perl… it hasn’t been updated in 2years? So let’s disclude that
from the discussion. If your Linux distribution is running an version
older then 5.8.4 (current is 5.8.8) you should be pretty much shot.

Unfortunately things change rapidly, things are updated, bug fixes
made. Some Linux distributions run on a 6-month Release Cycle, so in
6-months they play catchup and update all the things that they can.
They miss somethings, however that’s not here nor there. My knee-jerk
solution is if LAMP is so easy to deploy, then why not use it? or why
not use Perl or Python?

I admit I love the beauty of the Ruby Language, however if there’s
something out there that does the same job as Typo and is easier to
deploy and works “better” then I say go ahead.

It’s unfortunate that MarsEdit was one of your Hanging Chads… I
never got around to buying a License of MarsEdit, I just couldn’t be
bothered with it, I would rather write in TextMate and then copy and
paste.

However one of my long standing issues with Typo has been … drum
roll how I have to insert
's into my posts because I included a
html tag, and somehow it broke the whole mess. So either I can have
badly formatted posts that look worse then my emails … or I can
force a line break every here and there so make it look cleaner.

I do admit though that Frédéric responded slow on the Typo 5.0 Beta
(Rails 2) release to my dismay, I lost my typo install twice due to a
cache error, and it took several more people to have the same problem
before it was even brought up. But overall he try’s and that’s what
counts.

[chet]

On Wed, Jul 16, 2008 at 4:22 AM, Scott L. [email protected] wrote:

To whomever it may concern,
I notice the common thread here. How to deploy typo?

There is many ways to deploy typo, the most common is

1. FastCGI.

We all know, it sucks.

2. Mongrel/Webrick

Nginx with proxy load balancer to cluster of Thin/Ebb (+unix socket)
processes is a much faster.

3. Phusion Passenger (aka mod_rails?)

This is the best option. Fast and dead brain simple to deploy (just like
PHP).

Now, there’s no real difference with Mongrel/Webrick

Mongrel is much faster than Webrick. Thin is faster than Mongrel, and
Ebb is faster than Mongrel or Thin.

But those who have come from the “PHP Boat” (as we’ll call
it, a/k/a wordpress, etc) they just untar files into a directory edit a few
files, loadup their web browser and bam. It works.

Phusion mod_passenger (http://www.modrails.com/) works in the very
same way. Just copy your files, load web browser and bam. It works.
Need to reload application? No problem. Create empty restart.txt file
in the tmp folder. Apache will reload the application. No need to
restart Apache. No SSH needed. Simple FTP access is all you need.
mod_passenger works also with others Ruby frameworks which use
Rack,e.g. Merb. It even works also with Python…

You also forget mention Ruby Enterprise version which uses 33% less
memory… It works fine with mod_passenger and lowers the memory
footprint. http://www.rubyenterpriseedition.com/

mod_passenger is the easiest option for deployment Rails
applications (including Typo).

But, there is also another, 4-th deploy option: JRuby. It is also
brain dead simple to deploy. Just create WAR file and copy it into
webapps Tomcat’s folder. That’s all!

sudo jruby -S gem install warbler, activerecord-jdbc-adapter
jruby -S rails my_app

For developing use: jruby script/server; jruby script/console etc.

For production, just create ROOT.war file (with “warble war” command)
and copy it into webapps folder in Tomcat, Jetty or another Java
servlets container. This WAR file contains ALL gems (including JRuby
interpreter) so it works fine in all platforms.

[chet]

On Jul 16, 2008, at 3:47 PM, Chet F. wrote:

Application Y if both X and Y are competitors in the same market
(i.e., blogging software).

As I said above, you are doing nothing useful in this argument. Stop
it. If you care that much passionately, then run Wordpress. The
argument is self defeating, because you are comparing apples to
pears. You can try this argument on a Django list and see how far you
get.

there.

Why on earth would you need to run a second web server? That seems
like a really bad idea, frankly, hence my annoyance that the most
obvious question (which boils down to “WTF?”, essentially) isn’t
addressed.

Then I suggest you to take that torch up with the Mongrel Mailing list
and ask them.

staging or testing purposes).
3) copy the backed up databases (do not move them, we want to keep
a pristine copy still somewhere) to db/
4) verify you are using the proper environment (e.g. PRODUCTION /
DEVELOPMENT …) and then run rake db:migrate

What does rake do in this context?

Migrations.

To Quote from
http://wiki.rubyonrails.org/rails/pages/UnderstandingMigrations

"ActiveRecordMigration allows you to use Ruby to define changes to
your database schema, making it possible to use a version control
system to keep things synchronized with the actual code.
This has many uses, including:

 * Teams of developers – if one person makes a schema change, the

other developers just need to update, and run “rake db:migrate”.
* Production servers – run “rake db:migrate” when you roll out a
new release to bring the database up to date as well.
* Multiple machines – if you develop on both a desktop and a
laptop, or in more than one location, migrations can help you keep
them all synchronized."

In essence, your database has a ‘version’ number in it, and if someone
adds a new model, or changes a model a migration is also made. So
that you can run ‘rake db:migrate’ to ensure your database is “up to
date” and able to do what the new version intends to do.

To put it mildly, that’s a bit bizarre and very unfriendly to the
user.

With the exception of plugins and themes, I find the upgrade process
very relaxing and totally capable. Very straight forward, and easy to
do. You can stage an upgrade painlessly and if there’s a problem you
have not affected your actual blog in any way what so ever.

I believe as a standard practice

Maybe for Typo. Not for anything else I use.

Remind me not to hire you as a Systems Administrator. You should
ALWAYS make a backup, upgrade in a an test environment and verify
everything happened properly before you roll out the new version.
Now I realize that isn’t as “Cowboy” as say Wordpress, where you just
slap in the new version and it might upgrade your database and off you
go. But when things FAIL and they always do (see Murphy’s Law) it’s
better to have a backup on hand, and be able to recover from it. Then
fall flat on your ass and then resort to emailing someone saying “WTF
HOW DO I FIX THIS?”

Not to mention that is part of the Sarbanes-Oxley Compliance Matrix
(http://www.crcdataprotection.com/solutions/sox_compliance.asp
)

It’s always a good idea to have a backup handy, and there is a
wonderful list of things you have to do “to be Fully Sox-
Compliant” … I only mention this because some of these should be a
daily business practice for some people. It would make their life
easier.

[chet]

Scott,

You persist in answering questions that I’m not asking. At this point,
I won’t give a damn about Typo deployment again until some time
after the bug fixes I require are deployed, if then.

Your ongoing insistence that Typo/Ruby IS TOO! easy to set up and run
is, well, tiresome. Please stop.

disclude that from the discussion. If your Linux distribution is
running an version older then 5.8.4 (current is 5.8.8) you should be
pretty much shot.

Whatever.

Unfortunately things change rapidly, things are updated, bug fixes
made. Some Linux distributions run on a 6-month Release Cycle, so
in 6-months they play catchup and update all the things that they
can. They miss somethings, however that’s not here nor there. My
knee-jerk solution is if LAMP is so easy to deploy, then why not use
it? or why not use Perl or Python?

Seriously? You’re going there? “If (other thing) is so much better,
just USE IT!” Do you not realize how petulant that sounds? Are you
going to take your ball and go home now?

It’s unfortunate that MarsEdit was one of your Hanging Chads… I
never got around to buying a License of MarsEdit, I just couldn’t be
bothered with it, I would rather write in TextMate and then copy and
paste.

Post-from-local-client is a core requirement for me. Cut and paste is
for the birds. With Blosxom, I wrote in TextMate and saved. That’s it.
A cron’d rsync did my posting for me. With proper support for the XML-
PRC stuff, I can do that again with Typo; it works with WP and MT just
fine via anything that supports that interface, including MarsEdit and
TextMate via the blogging bundle.

However one of my long standing issues with Typo has been … drum
roll how I have to insert
's into my posts because I included a
html tag, and somehow it broke the whole mess. So either I can have
badly formatted posts that look worse then my emails … or I can
force a line break every here and there so make it look cleaner.

There are lots of ways to solve this. I post in Markdown.

I do admit though that Frédéric responded slow on the Typo 5.0 Beta
(Rails 2) release to my dismay, I lost my typo install twice due to
a cache error, and it took several more people to have the same
problem before it was even brought up. But overall he try’s and
that’s what counts.

Try that statement at work and see how far you get.

Don’t get me wrong; I appreciate Frederic’s answers very much – and,
not to put too fine a point on it, his English is better than yours –
but eventually I’ll make a choice based on function and support, not
how hard the maintainers are trying.

---

“They say no mortal woman was enough for him so he made one himself
outta whiskey an liquors an ale,” says me. “An he loved her like a
lumberjack made of eating loves a woman made of ham.” (Fafblog
2004-08-05)

[chet]

On Jul 16, 2008, at 3:36 PM, Chet F. wrote:

Unfortunately things change rapidly, things are updated, bug fixes
made. Some Linux distributions run on a 6-month Release Cycle, so
in 6-months they play catchup and update all the things that they
can. They miss somethings, however that’s not here nor there. My
knee-jerk solution is if LAMP is so easy to deploy, then why not
use it? or why not use Perl or Python?

Seriously? You’re going there? “If (other thing) is so much better,
just USE IT!” Do you not realize how petulant that sounds? Are you
going to take your ball and go home now?

It’s not my ball, it’s everyone else’s ball. I do not suffer the
problem of a 6-month release cycle, or how painful Ruby on Rails is to
deploy. I realize I don’t see your pain, so perhaps if you could
actually write up a way we could make this easier for you? Because
clearly I don’t get it, and won’t get it. So instead of venting your
frustration the way you are, please educate us.

including MarsEdit and TextMate via the blogging bundle.
Sounds like an excellent feature to be proposed for Typo then. The
proper XML-RPC Support so you can post from TextMate.

However one of my long standing issues with Typo has been … drum
roll how I have to insert
's into my posts because I included
a html tag, and somehow it broke the whole mess. So either I can
have badly formatted posts that look worse then my emails … or I
can force a line break every here and there so make it look cleaner.

There are lots of ways to solve this. I post in Markdown.

I use Textile and I guess that’s why I have that problem.

yours – but eventually I’ll make a choice based on function and
support, not how hard the maintainers are trying.

Then I will stop bothering now, because it’s clear that I am wasting
my time.

[chet]

On Jul 16, 2008, at 5:09 PM, JZ wrote:

On Wed, Jul 16, 2008 at 4:22 AM, Scott L. [email protected] wrote:

To whomever it may concern,
I notice the common thread here. How to deploy typo?

There is many ways to deploy typo, the most common is

1. FastCGI.

We all know, it sucks.

It sucks, but “it takes less memory” to quote someone I know who uses
Lighttpd + FastCGI. His rather old RSS Reading application took 250
Megabytes of memory, for which is “not there” in FastCGI Mode.

2. Mongrel/Webrick

Nginx with proxy load balancer to cluster of Thin/Ebb (+unix socket)
processes is a much faster.

If thin supported Streaming (Not Media, but the Mongrel Extension) it
would be something I would have no issue using. I’ve mentioned that
to the author of Thin and we’ll see how that goes.

But those who have come from the “PHP Boat” (as we’ll call
Rack,e.g. Merb. It even works also with Python…
webapps Tomcat’s folder. That’s all!

sudo jruby -S gem install warbler, activerecord-jdbc-adapter
jruby -S rails my_app

For developing use: jruby script/server; jruby script/console etc.

For production, just create ROOT.war file (with “warble war” command)
and copy it into webapps folder in Tomcat, Jetty or another Java
servlets container. This WAR file contains ALL gems (including JRuby
interpreter) so it works fine in all platforms.

I know a few people who have been deploying Merb apps in Tomcat that
way. I understand it’s a very easy way to pass off a Merb app as a
“Java” Webapp without someone knowing what it really is. As far as
Ruby Enterprise Version I was not aware of that, but like everything
you don’t always find all the options in 1 concise web page.

[chet]

On Wed, Jul 16, 2008 at 4:47 PM, Chet F. [email protected] wrote:

Why on earth would you need to run a second web server? That seems like a
really bad idea, frankly, hence my annoyance that the most obvious question
(which boils down to “WTF?”, essentially) isn’t addressed.

Let’s compare to a Wordpress blog run on Apache using mod_php. The PHP
code does not serves all requests. Apache serves up the static content
(css, javascript, graphics, etc.). Many popular front-end servers also
perform proxying and load balancing, and tons of other functions.
Mod_php just runs the PHP code. Most Ruby web apps are served in this
fashion rather than by an Apache module because it was easier to
implement (at the time mod_ruby was a memory-leaking nightmare) and it
is a more flexible design by virtue of not being tied to Apache and by
being proxyable behind any decent http proxy.

It’s not a bad idea, it’s a sound idea. Please settle down and ask
polite questions if you truly want to understand and use Typo.

[chet]

On Thu, 17 Jul 2008, JZ wrote:

On Wed, Jul 16, 2008 at 4:22 AM, Scott L. [email protected] wrote:

2. Mongrel/Webrick

Nginx with proxy load balancer to cluster of Thin/Ebb (+unix socket)
processes is a much faster.

A bit overkill for 99.9% of all blogs, though.

[chet]

On Jul 16, 2008, at 7:34 PM, Kevin W. wrote:

(css, javascript, graphics, etc.). Many popular front-end servers also
perform proxying and load balancing, and tons of other functions.
Mod_php just runs the PHP code. Most Ruby web apps are served in this
fashion rather than by an Apache module because it was easier to
implement

This must be some radical other definition of “easier.”

(at the time mod_ruby was a memory-leaking nightmare) and it
is a more flexible design by virtue of not being tied to Apache and by
being proxyable behind any decent http proxy.

It’s not a bad idea, it’s a sound idea. Please settle down and ask
polite questions if you truly want to understand and use Typo.

If you really think I’m the one that needs to settle down, I think you
must have been reading some other thread.

I’ve made it abundantly clear that I am 100% uninterested in
installation tweaks right now because I am dealing with bugs in Typo.
That’s my front-burner issue. If these bugs can’t get resolved, the
installation issues are irrelevant. I appreciate you explaining the
rationale for Mongrel, et. al., above, nevertheless.

Chet

“An educated public is exactly what they don’t want . . . more people
believe in angels than believe in long division.” R. Norris 31 May 2006

[chet]

On Wed, Jul 16, 2008 at 6:22 PM, Scott L. [email protected] wrote:

If thin supported Streaming (Not Media, but the Mongrel Extension) it would
be something I would have no issue using. I’ve mentioned that to the author
of Thin and we’ll see how that goes.

Thin depends on the eventmachine library, which is not great for
uploads because it is more of a fire-and-forget event-driven
architecture and long-running requests mess it up. Mongrel is better
at uploads. There is a solution for this for Merb users:
http://brainspl.at/articles/2008/04/18/deferred-requests-with-merb-ebb-and-thin
I have no idea if something like this can be used in a Rails app, but
that would be up to Rails/Typo to make that happen.

[chet]

On 16 Jul 2008, at 14:38, Chet F. wrote:

that this is actually driving me to think I should can it and go
for the same suite of PHP apps as everyone else.

I will agree with that, as Debian Etch currently has Ruby 1.8.4(2?
i forget) with Rubygems 0.92. However is that Ruby’s problem? or
the Linux distribution you chose?

It’s definitely Ruby’s problem if PHP, Perl, Python, etc., are all
running fine out of the box.

I’d just like to put in a vote for not Ruby’s problem here. I’ve never
had any trouble deploying rails applications. I used to be a mod_perl
hacker and that was much, much harder to set up and keep going.

The only difference, in my view, is that Rails isn’t a commodity
solution yet. You can’t easily buy some Rails and you don’t get an
option on the Ubuntu disc to install a good starting Rails setup. A
Rails app needs a port, I suppose, so you can’t really run one unless
you have your own box and it’s really not something you should bother
with if you just want your blog to be fashionably served.

If you have some reason to want Apache as your front end, you have to
know how to proxy to another port. The documentation for that is here:

mod_proxy - Apache HTTP Server Version 2.2

and includes straightforward cut and paste configuration along with
some very useful warnings. I don’t know anything about mod_rails, but
I suspect that unless you want to get fancy with the apache lifecycle,
you don’t need that much integration. Nginx is a much better front end
anyway: fast and simple. There’s an excellent cargo config here:

http://brainspl.at/articles/2006/09/12/new-nginx-conf-with-rails-caching

and some thorough benchmarking here:

Looking For Optimal Solution: Benchmark Results Summary and Findings :: Oleksiy Kovyrin

I’ve found it perfectly straightforward to set up typo (or radiant, or
mephisto: I have sites running on each) using mongrel_cluster,
capistrano and an nginx front end. The only things I had to compile
were nginx and sphinx. Everything else is apt-gettable (and I think
now nginx is too). I use three application servers and one database
server and deliver over 100,000 pages a day with typically about a
quarter load. It scales well enough for me and it’s over two years
since the last boot. I certainly couldn’t say that when I was
desperately propping up 100MB apache processes.

Here, you’re defaulting back to a knee-jerk defense of what is
clearly your pet language. That has no place here. Compared to LAMP-
stack stuff, RoR applications are much harder to set up and deploy.
They require a totally different approach, and that approach is very
poorly documented. This isn’t a controversial statement.

The documentation is fine. The only problem is that there is no single
orthodox solution. I see that as a strength, but it does mean that
some expertise is required to choose your recipe. You (Chet) are right
in the sense that for the beginner, a working typo blog is probably
not as easy to get to as a working php-based blog. For anyone who
knows what they’re doing there’s really no difference and the rails
model is much easier to maintain.

Most of this is general to rails so it’s also worth mentioning that
Frédéric is very diligent and responsive and the software is good. He
deserves more appreciation, i think.

best,

will

[chet]

He deserves more appreciation, i think.
Regarding the dcumentation, I’ve started to refactor the one we have
and published some. I’m taking the doc back on the redmin, and will
move *.typosphere.org on it but the blog.

If someone here have the time to proofread this :
http://redmine.typosphere.org/wiki/typo/Before_Installing_Typo
http://redmine.typosphere.org/wiki/typo/Install_Typo_with_Typo_installer
http://redmine.typosphere.org/wiki/typo/Install_Typo_from_the_gem
http://redmine.typosphere.org/wiki/typo/Secure_Typo_admin_with_HTTPS

I will add docs on deploying Typo under various things tonight :
– Apache + mod_rails
– Apache + fastcgi
– Apache + mongrel
– Nginx + Thin
– Nginx + Fastcgi
– Nginx + mongrel
– Lighttpd + fastcgi
– Lighttpd + Thin
– Lighttpd + mongrel

However, we do officially recommend and support mod_rails as the best
and easiest way to run any Rails app.

Regarding my appreciation, I know I’ve made some mistakes (5.0 was one
of them : relying too much on the tests and not testing in production
mode enough). However, I prefer having negative feedback as long as
it’s argumented and documented, it’s much better than just “yay, your
appz is so cool” when you want to improve things.

Regards,
Frédéric

–
Frédéric de Villamil
[email protected] tel: +33 (0)6 62 19 1337
http://fredericdevillamil.com Typo : http://typosphere.org

[chet]

On Jul 16, 2008, at 7:12 PM, Scott L. wrote:

Stop it. If you care that much passionately, then run Wordpress.
The argument is self defeating, because you are comparing apples to
pears. You can try this argument on a Django list and see how far
you get.

The problem here, Scott, is that you’re the one who’s arguing. I
stated an uncontroversial position – RoR apps are harder to deploy
than traditional LAMP apps – and you’ve gone all apoplectic with
fanboy protestations that, frankly, make no sense.

Look: I don’t care. It’s absolutely not important to me today, and I’m
100% done with you. I have problems with Typo that Fred says he’s
working on, and I appreciate that. What I do NOT appreciate is your
incessant browbeating and bombastic Ruby boosterism.

Why on earth would you need to run a second web server? That seems
like a really bad idea, frankly, hence my annoyance that the most
obvious question (which boils down to “WTF?”, essentially) isn’t
addressed.

Then I suggest you to take that torch up with the Mongrel Mailing
list and ask them.

If it’s something that’s considered a common part of a Typo install,
then the Typo docs need to address it because it’s out of the ordinary
for weblogging software.

To Quote from http://wiki.rubyonrails.org/rails/pages/UnderstandingMigrations

I’ll read this later. I am refreshed, however, that you’ve provided an
apparently useful link.

In essence, your database has a ‘version’ number in it, and if
someone adds a new model, or changes a model a migration is also
made. So that you can run ‘rake db:migrate’ to ensure your database
is “up to date” and able to do what the new version intends to do.

Why is this better than just dropping the SQLite file into the tree
after your re-install? (Hypothetically; I haven’t tried it.)

Of course you’ll have to reinstall your plugins and themes.

!!!

To put it mildly, that’s a bit bizarre and very unfriendly to the
user.

With the exception of plugins and themes, I find the upgrade process
very relaxing and totally capable.

It’s plugins and theme reinstallation I find bizarre.

I believe as a standard practice

Maybe for Typo. Not for anything else I use.

Remind me not to hire you as a Systems Administrator.

Based on my exposure to you here, it seems astoundingly unlikely
you’ll ever be in a position to hire anyone with my resume.

My reference here is to the need to reinstall plugins and themes, not
standard pre-patch/pre-upgrade backups. In my career so far, my
experience is much more defined by my own refusal to hire doctrinaire
platform zealots.

Not to mention that is part of the Sarbanes-Oxley Compliance Matrix (http://www.crcdataprotection.com/solutions/sox_compliance.asp
)

SarBox is so often a part of blog culture.

Because clearly I don’t get it, and won’t get it.

This, at least, is abundantly clear.

Chet F.

“When you’ve got an RV, a jet pack, and a monkey you really don’t need
much actual content” - KS

[chet]

On Wed, Jul 16, 2008 at 11:21 PM, Chet F. [email protected] wrote:

On Jul 16, 2008, at 7:34 PM, Kevin W. wrote:

Let’s compare to a Wordpress blog run on Apache using mod_php. The PHP
code does not serves all requests. Apache serves up the static content
(css, javascript, graphics, etc.). Many popular front-end servers also
perform proxying and load balancing, and tons of other functions.
Mod_php just runs the PHP code. Most Ruby web apps are served in this
fashion rather than by an Apache module because it was easier to
implement

This must be some radical other definition of “easier.”

At the time Mongrel came around, no one, and I mean no one, would
touch the mod_ruby code. Webrick was slow and FastCGI was very buggy
at best. Zed S. stepped up with a strictly-spec-compliant and fast
web server for Ruby. There was much rejoicing and dancing in the
streets. You can call it radical if you want to, but it worked very
well.

(at the time mod_ruby was a memory-leaking nightmare) and it
is a more flexible design by virtue of not being tied to Apache and by
being proxyable behind any decent http proxy.

It’s not a bad idea, it’s a sound idea. Please settle down and ask
polite questions if you truly want to understand and use Typo.

If you really think I’m the one that needs to settle down, I think you must
have been reading some other thread.

I politely asked you to change the tone of your requests, yet you
continue to be confrontational. Please stop. Comments such as “that’s
ridiculous” and “that’s a very bad idea” and repeatedly saying “you’re
wrong” when you’ve clearly shown that you don’t understand why things
are the way they are now is just causing trouble. You have issues with
Typo and want help. Is this really how you ask for help?

I’ve made it abundantly clear that I am 100% uninterested in installation
tweaks right now because I am dealing with bugs in Typo. That’s my
front-burner issue. If these bugs can’t get resolved, the installation
issues are irrelevant. I appreciate you explaining the rationale for
Mongrel, et. al., above, nevertheless.

You’re welcome, and thank you for saying so.

[chet]

On Jul 17, 2008, at 10:02 AM, Kevin W. wrote:

At the time Mongrel came around, no one, and I mean no one, would
touch the mod_ruby code. Webrick was slow and FastCGI was very buggy
at best. Zed S. stepped up with a strictly-spec-compliant and fast
web server for Ruby. There was much rejoicing and dancing in the
streets. You can call it radical if you want to, but it worked very
well.

I appreciate you giving the backstory here, but having to include
another web server, though, is not something I’d describe as “easier”
when compared to other tools that require only one.

Please settle down and ask
polite questions if you truly want to understand and use Typo.

If you really think I’m the one that needs to settle down, I think
you must
have been reading some other thread.

I politely asked you to change the tone of your requests, yet you
continue to be confrontational.

I was confrontational to Scott because his posts were snide, rude,
useless knee-jerk defenses of his pet stack.

Please stop. Comments such as “that’s
ridiculous” and “that’s a very bad idea” and repeatedly saying “you’re
wrong” when you’ve clearly shown that you don’t understand why things
are the way they are now is just causing trouble.

My complaints really began with this very point; recall one of my
original points was the poor state of documentation concerning the
idiosyncratic installation requirements of Typo.

If it were more clear why I should consider installing another web
server just to run Typo from Typo’s own docs, perhaps I wouldn’t be
complaining about the docs.

You have issues with
Typo and want help. Is this really how you ask for help?

I certainly wasn’t asking Scott for help, and he certainly wasn’t
providing any. He jumped in to tell me my assessment of the state of
Ruby/Rails/Typo installation was wrong, which is simply incorrect.
He’s offered nothing of value to me in re: my actual problems.

My original posts to this list met with either no response at all, or,
more recently, helpful responses from Frederic. Frankly, I was
surprised, after that, to get Scott’s fanboy eruptions. I believe I
dealt with him appropriately, and with a level of grace appropriate
for handling such a poster.

Chet

[chet]

On Jul 17, 2008, at 11:36 AM, Chet F. wrote:

I certainly wasn’t asking Scott for help, and he certainly wasn’t
providing any. He jumped in to tell me my assessment of the state of
Ruby/Rails/Typo installation was wrong, which is simply incorrect.
He’s offered nothing of value to me in re: my actual problems.

My original posts to this list met with either no response at all,
or, more recently, helpful responses from Frederic. Frankly, I was
surprised, after that, to get Scott’s fanboy eruptions. I believe I
dealt with him appropriately, and with a level of grace appropriate
for handling such a poster.

Perhaps you mistook my emails as a knee-jerk nonsense emails. But I
was trying to get more information to provide some assistance, however
I was not getting any valid feedback. I realize I did not word them
in a way that you wanted, however my intent was to get more
information to help.

I searched through old emails trying to find any information of why
you could not do this or that, and Truthfully all I could find were
posts related to “MarsEdit”. Which those bugs have been fixed in
trunk, if you want you can try updating to “Trunk” and see if that
fixes your problems?