It’s just data

FeedTools on Ubuntu

When I first heard about FeedTools, I was skeptical.  I’ve seen many attempts at writing feed parsers by authors who have been seduced by the claims that that feeds are simple; in most cases these parsers end up being abandoned before they ever really become complete.

Loooking again, it appears to me that Bob Aman has the persistence necessary to take this task to completion, as such FeedTools merits a second look.  I’m installing from tar files as I expect to make patches.  In particular, the application I’m considering, I need to maintain the distinction between summaries and content.

To understand why, consider the feeds of Mena Trott or Tim Bray.  If I find a link to my weblog in the full content, I want tidy and trim the shorter summary as the excerpt.  Of course, in feeds where both are not provided, I’ll take what I can get.  Effectively this means that I scan the (content or summary) and then tidy and trim the (summary or content) of the ones that match.

In any case, the installation on Ubuntu turned out to be harder than I expected, but here is what I came up with:

Add Universe.

sudo apt-get install ruby rdoc
 
wget http://rubyforge.org/frs/download.php/5207/rubygems-0.8.11.tgz
tar xzf rubygems-0.8.11.tgz
cd rubygems-0.8.11
sudo ruby setup.rb
cd ..
 
sudo gem install -y rake activerecord
 
sudo apt-get install libnet-ssh-ruby1.8 libtidy0
sudo gem install uuidtools tidy
 
sudo apt-get install build-essential ruby1.8-dev
sudo apt-get install sqlite libsqlite-dev
sudo gem install sqlite-ruby
2
 
wget http://rubyforge.org/frs/download.php/6532/feedtools-0.2.16.tgz
tar xzf feedtools-0.2.16.tgz
cd feedtools-0.2.16
sqlite test.db < db/schema.sqlite.sql
cat <<EOF >database.yml
production:
  adapter: sqlite
  dbfile: test.db
EOF
 
rake

Oh, and one request: in the future, can all of the ._ files be stripped from the archives?


Sam Ruby: FeedTools on Ubuntu

[link]...

Excerpt from del.icio.us/alpuckan/ubuntu at

Neat, so I guess that means sqlite works?  I never managed to successfully get sqlite installed on my PowerBook, so sqlite support has actually been a somewhat untested feature.  Most of my testing is done against PostgreSQL.

And yeah, somewhere along the way I made a comment that RSS is probably one of the most horribly misrepresentative acronyms ever.  It’s anything but simple.

There’s no doubt that FeedTools is a long way from being done, and realistically I suppose it never really will be done, but yeah, I’m gonna see it through.

I’m a little confused by your “request” though.  What ._ files?

When I have time, I do plan on coming up with a decent way of letting people have their cake and eat it too with the summaries/content issue.

Posted by Bob Aman at

There’s no doubt that FeedTools is a long way from being done, and realistically I suppose it never really will be done, but yeah, I’m gonna see it through.

IMHO, the best way to proceed is to more completely leverage the UFP’s test suite.

I’m a little confused by your “request” though.  What ._ files?

Try the following to see what I mean:

tar tvf feedtools-0.2.16.tgz | grep "\._"

When I have time, I do plan on coming up with a decent way of letting people have their cake and eat it too with the summaries/content issue.

It’s pretty easy.  Define summary and content separately.  The search path that each looks through can overlap completely, partially, or even be disjoint.  Then define description to be either content or summary or summary or content, possibly based on a configuration option.  Edit a Bloglines subscription, and see “Display Preferences” to see this in action.

Posted by Sam Ruby at

Well, I’d really, really like to avoid configuration stuff related to how the feed data is presented.  Configuration related to whether to enable tidy, or which caching mechanism to use — that’s one thing, but “should the description method return summaries or full content?” — that’s another ball of wax entirely, IMHO, and it’s something I’d rather avoid.  Again, stylistic, but I tend to feel it’s the more ruby-like way of doing it.

I assume you meant: tar tzf feedtools-0.2.16.tgz | grep "\._"

Interesting, but...

bob@sporkmonger:~/Desktop $ tar xzf feedtools-0.2.16.tgz
bob@sporkmonger:~/Desktop $ cd feedtools-0.2.16
bob@sporkmonger:~/Desktop/feedtools-0.2.16 $ ls -al
total 56
drwxr-xr-x    9 bob  bob   306 Oct 18 16:16 .
drwx------   66 bob  bob  2244 Oct 31 13:58 ..
-rw-r--r--    1 bob  bob  4713 Oct 18 15:35 CHANGELOG
-rw-r--r--    1 bob  bob   462 Oct 31 13:58 README
drwxr-xr-x    5 bob  bob   170 Oct 31 13:58 db
-rw-r--r--    1 bob  bob   741 May 29 10:47 install.rb
drwxr-xr-x    4 bob  bob   136 Oct 31 13:58 lib
-rw-r--r--    1 bob  bob  3046 Oct 31 13:58 rakefile
drwxr-xr-x    9 bob  bob   306 Oct 18 16:16 test
bob@sporkmonger:~/Desktop/feedtools-0.2.16 $ cat ._CHANGELOG
cat: ._CHANGELOG: No such file or directory

Also, I’m using rake (see rakefile) to generate the tarballs/zips/gems, so if there actually is a problem, it’s probably either in there or on your end somehow.  I can’t seem to get tar to actually extract the ._ files to my filesystem, and the only way I can even tell they’re there is with tar tzf feedtools-0.2.16.tgz | grep "\._".

Posted by Bob Aman at

Well, I’d really, really like to avoid configuration stuff related to how the feed data is presented.  Configuration related to whether to enable tidy, or which caching mechanism to use — that’s one thing, but “should the description method return summaries or full content?” — that’s another ball of wax entirely, IMHO, and it’s something I’d rather avoid.  Again, stylistic, but I tend to feel it’s the more ruby-like way of doing it.

OK.  Perhaps some sort of hybrid model would work.  Organizing similar tests into directories.  Having a small ruby driver which sets up options on a per directory basis, and then simply iterates over all of the files in the directory.

I assume you meant: tar tzf

Yes.  Oddly, tvf works on my machine.  In any case, peeking inside these files, it appears that they were created by something called /Applications/TextMate.app on Mac OS X.

Posted by Sam Ruby at

Well, that is the text editor I tend to use.  But I can’t, for the life of me, get OS X to admit that they exist.

I’ll see if I can’t just get tar to do the deleting after the fact.

Posted by Bob Aman at

I finally figured out what was creating those ._ files.  Apple has a magical version of tar that includes the metadata files.  Terribly retarded.  Unfortunately, I don’t have the time to recompile tar myself and replace Apple’s version at the moment, so you’ll have to suffer though at least one more release with the stupid ._ files, but I’ll make sure they’re gone by the next version.

Posted by Bob Aman at

Comment on You mean it's not vaporware? by Sam Ruby

About three months ago we discussed what it would take to provide separate summary and content… did this drop off of the list?...

Excerpt from Sporkmonger: You mean it's not vaporware? at

Add your comment