Its the holidays and I am having a whale of a time relaxing at home, watching brainless american pulp. I had to migrate some of my mail servers so I thought it would be fun to add on Greylisting to all my MTA's. Turns out Greylisting was dead easy setup and now I'm left twiddling my thumbs looking for something else to do … inbetween all the anime, tv and gaming.
WebMail I thought to myself, always wanted it might as well do it!
So here are the candidates:
- Squirrelmail - The granddaddy, 10 years and counting… this was around when I first started playing with linux
- Roundcube - Nice and purrty ...
- Bluemamba - With TMDA support
- DIMP - addon to the Horde, ajaxy version of their IMP, also interesting is their MIMP project made for mobiles…
- Ilohamail - Seems like its no longer a running project
- Uebimiau - demo instances are down :( It looked kinda cool…
NB: there are a lot more packages I have not included since I am looking for something which only does webmail and plays nicely with Postfix & Dovecot. A few of the noteable excluded packages are Zimbra, Scalix and Axigen. Most of these are too heavy for what I want this time, usually they are part of a larger solution with its own backend services and heavy.
Don't Call Me Shallow!
Let it be said I am shallow, I like things to look slick… which is why i'm on a mac… Here are some screenshots of the the remaining 4 contestants:
DIMP + IMP + Horde
The only notable one here is Squirrelmail for its fuglyness… :( lets hope I don't need any special features it has.
A House Needs Rooms!
Lets see if any special features which would be nice to have… Some of the requirements I have forgotten to mention up till now are:
- Linux based, preferably LAMP, but i'm flexible on the 'M' part and the 'P' part.
- Preferably already inside debian/ubuntu repo (not a deal breaker)
- Light weight
- Scalable in both performance and maintenance
- A project with momentum, enough support to debug those rare issues
- Plays well with postfix, dovecot… (most of these webmail seems to talk through POP/IMAP which on the surface seems strange to me, seems to sacrifice efficiency for compatibility…
- Needs to be purrty….
Another one bites the dust… took a quick look at the BlueMamba site, it is a relatively young project (15 months). Bluemamba doesn't have much documentation online nor support. All the default Sourceforge tracker, forums are empty.
BUGS uh features!
- Horde + IMP + DIMP
The main reason for this order is because of the scope of my needs. Right now I just need the ability to view my emails when I am not in front of my own computer(s). E.g. no access to Apple Mail… like travelling or internet cafe's. Since I'm shallow Roundcube makes the most sense. In a larger corporate environment I would more than likely look at Squirrelmail or Horde first. Squirrel's plugin system makes it a very attractive choice, its a stable mature project deployed in production environments and proven to scale. Horde very much sits on the same level as Squirrel in-terms of its connectivity with the bonus of a nice Ajax front-end in DIMP.
Getting my hands dirty installing these software will likely reveal a lot more about the internals of these software and could drastically change my opinion of these packages. You can tell a lot about a project by its installation process…
One of the things I am looking at these days is a simple to manage "Single Sign On" (SSO) solution. The direction I want to first have a look at is doing this at the Apache level. It needs to talk to LDAP and RDBMS. Quick breakdown:
SSO which sits at the apache level:
Some of the standalone SSO solutions:
- Apache Shiro homepage
- Atlassian Crowd
- CoSign - University of Michigan
- Pubcookie - University of Washington
Some of the SSO implementations which are not a good fit for me:
I just stumbled upon a great lecture at Stanford university by a successful entrepreneur Eric Ries, co-founder of IMVU.
He talks about how to succeed and fail at a start-up. Retelling some of the mistakes along the way.
Some highlights include:
- Short iteration cycles, don't build big software that no one will use.
- learn fast and if something's not working change direction, something he calls the "pivot"
- Waterfall should be avoided unless both the problem and the solution is very well known (which is almost never)
- You think you know what customers want but you don't because you haven't asked
- All major assumptions (what he calls the shadow beliefs) should come out in the open.
Highly recommended to anyone who has started, thinking about starting or just thinking about a start-up.