TODO list for 1.1.6 release

* TODO list for 1.1.6 release
@ 2009-07-08 12:23 Luis Lavena
       [not found] ` <71166b3b0907080523r458a237btb6edb2ae9c219a72-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
  0 siblings, 1 reply; 12+ messages in thread
From: Luis Lavena @ 2009-07-08 12:23 UTC (permalink / raw)
  To: Mongrel Developers (ML)

Hey Evan and others watching this list.

I would love to release an official small point release for Mongrel in
the next couple of days. Mainly, the following things is what I'm
aiming:

* Basic Ruby 1.9 functionality

This means "it just built and run", it doesn't mean anything optimized
for 1.9. Bigger versions can address those details.

There is a weird Zlib error I'm going to investigate.

* Modernize infrastructure

- Latest Ruby versions do not require fastthread, since has been
integrated by default (1.8.6 greater than p114)
- CGI has been patched since 1.8.6-p114

I believe those 2 dependencies can be dropped, and
required_ruby_version of Mongrel itself bumped to 1.8.6 instead of
1.8.5, since due our changes in API compatibility may not build.

Also, I've invested a couple of months creating rake-compiler:

http://github.com/luislavena/rake-compiler

Not a shameless self-promotion, but it has been working great for
other projects and also removed the need to depend or require a
Windows building environment to be able to release Windows gems.

Add to that the concept of "fat binaries", and you get lot of happy
Windows users on Ruby 1.8 or 1.9 :-)

* Faster release cycles.

Let's get serious. There was no release of Mongrel itself in the past
YEAR. We all know Mongrel *is* stable, but things can get obsolete
pretty easily in the software industry.

I still use Mongrel on Windows for development and still deploy to
Linux servers running it, or Windows intranets using mongrel_service.

While some people had migrated to Passenger, there is still lot of
value on Mongrel itself.

So, ranting on a side, let's take the advantage of Git and GitHub,
provide a working stable branch/master and start looking at quality
patches.

That is the purpose of the branch "spinoff" available in my fork:

http://github.com/luislavena/mongrel/tree/spinoff

Things that are missing there:

* Steps to build the Java gem

rake-compiler hasn't been extended to get Java extensions build, been
waiting for some patches for that but no luck, so going to add a
tasks/java.rake for now.

* Ensure all tests pass, on all platforms.

There are some weird tests that I just skipped on Windows, need
further testing to see what is wrong with them, which is going to take
me some time.
Except from that and the 1.9 error described at the top of this email,
everything passes on OSX and Linux.

Something really weird I noticed is on threaded tests. They work
differently with 1.8.6-p287 and 1.8.6-p368, I believe this is the main
difference on the stalled tests.

Going to do a multiruby pass and see the results (stay tuned).

So, mainly, that is my reason for the spinoff branch, what do you guys think?

Evan: I know you forked this branch, and committed some patches from
mattetti, but while you see the value of those (specially one huge
commit) I see a mess to trace back "what the heck" something was not
working anymore.

Let's keep this conversation and see if we can address those issues. I
have a few days off from work that I want to take advantage of.

Cheers,
-- 
Luis Lavena
AREA 17
-
Perfection in design is achieved not when there is nothing more to add,
but rather when there is nothing more to take away.
Antoine de Saint-Exupéry

^ permalink raw reply	[flat|nested] 12+ messages in thread