Most obvious answer would be "don't": there are already lots of webmail software, some of which are PHP-based (if you depend on using PHP, because you already have a server based on a LAMP stack, for instance).
To quote only a few names, all PHP-based, you could have a look at these ones:
- SquirrelMail: a quite old-one, and not really sexy... But has been doing the job for years
- Horde IMP: well-know too, and quite powerful
- roundcube: maybe the most "sexy" and "web 2.0" of the list of those I tried
- AtMail: I've never used this one, so I can't say much more...
I've used the three firsts of the list for quite some time ; roundcube was definitely the most "user-friendly", I'd say...
(Googling a bit, you might find many more -- but I think I spoke about the ones that are the most used)
Now, if you have to set up a Webmail, say, for you company: definitely use some already existing software:
- an existing software would be quite "good" already:
- many people using it, which means many people who said "this could be done better", or "that is not user-friendly", or... you get the point ; all this made the existing software better :-)
- many people will have tried to break such an application -- which means lots of security-fixes, which means an application probably more secure than you'll write in a long time...
- an existing open-source application will represent hundreds of days of work
- are you really ready to spend that kind of amount of time working on something that already exists?
- if you're working for a company: is your company ready to pay your for one year? or even probably more ? working on something that already exists?
- you don't believe me? Read this blog-post: Code: It's Trivial ; it's about Stack Overflow, but it would be exactly the same for a webmail software (except that Stack Overflow is more recent -- and probably had less code-contributors ; but that's just a wild guess ^^ )
As a sidenote: if your company wants you to build a clone of gmail, you won't (same thing: how many people worked on gmail? How many programmers does your company have? How much can your company spend on this? )...
And for just a couple dollars each year, your company could have a "professionnal" Google account for each one of it's employes, BTW...
After all this, if you still want to / have to write a custom hand-made webmail using a LA*(M)*P stack, you will need to know at least the following:
- How to programm in PHP
- The basics of IMAP (even if you use a library / framework, such a big application will require you to know some stuff about the underlying protocols, or won't ever understand "why" or "how" something went wrong)
- HTML / CSS / JavaScript (with some knowledge of AJAX -- come on, it's 2009 !)
- So your application is easier to maintain (and adding/reworking/modifying stuff and correcting bugs is possible), you will probably want to use some Framework, and follow some design patterns and best practices
- As an example, you can take a look at Frameworks like symfony, or Zend Framework
- Yes, you can develop and application without a Framework, and/or without MVC... But.. ergh...
Considering all this is not a problem (If you are not quite good at all this already, and/or don't have much experience, it could take at least a couple of years to acquire that... Considering programming and web-developping is your full-time activity), you can start tkinking about accessing a mail server using, for instance, IMAP.
There are several possibilities here.
I would have a look, at least, before choosing, at these two:
Once your application is quite done, you will (hopefully !) start getting users, which means at least three things:
- Bug-reports ; those are almost always urgent for users -- like "OMG I don't understand nothing works anymore, I gotta get my mail RIGHT NOW!"
- Requests for evolutions: users always want more stuff like "I saw that in gmail on my personnal account ; how is it we don't have that too with our corporate account? It's the tool we use to work !"
- "This is too slow, I can't work !" ; then, pray you thought about scalability and optimization when you designed/developped the application!
Here, again, are you ready to deal with that?
Well, I think I said enough ; now, it's your time to think: do you really want/need to develop such an application from scratch?
If you have a bit of free time, maybe you could participate in an already existing, open-source, project ? That could be profitable to every one ;-)
Finally, one last thing: **if you want to work on such a project just for fun, to know what it's like, and to learn more about web-development, then DO !**
*(It's, in my opinion, probably the only reason that would justify working on this, BTW)*