jump starting a brand new website (under the gun)

[hlslaughter]

i've got a client who wants to see some immediate results on a brand new website. within a week, they'd like to see the following:

1) basic user authentication (using php sessions/cookies to maintain authentication)
2) some basic content (text, photos, audio) uploading mechanism
3) basic content browsing in a somewhat neatly formatted presentation

knowing how quickly they wanted something they could look at, i initially suggested using a framework such as mambo or seagull, but they don't want to have to deal with the scalability issues and other restrictions of those types of tools (issue is closed at this point, so no need to argue scalability of frameworks 🙂

oh, they also want a db abstraction layer so as to allow for future flexibility and the ability to use a distributed db backend down the road. i'm assumming i'd be writing this using adodb lib.

i'm a little nervous promising all this in a week.

i've got two questions for the esteemed audience members here:

1) does this seem like a reasonable goal for a one week timeframe?

2) is there any sort of "mini framework" that is not a full fledged graphically oriented content management system but rather a "starter kit" for a website that might have modules for stuff like authentication and session management, file uploads, maybe even a bit of content management, and other types of tasks that nearly all websites have in common? i'm talking something that would be flexible enough to allow for db abstraction, something that would not assume the role of the website's primary purpose but assume a sort of background role that allows a developer to skip a lot of the tedius coding tasks.

all the tools i've looked at (and some of them are indeed very nice) seem oriented towards non-programmers. that is, they assume that your going to be using their graphical interface to do most (not all) of your website creation.

in addition to db abstraction (using adodb), i would like to be using the best MVC framework that i'm aware of for php, smarty.

this is a one man show, and i'd really like to meet the customer's needs.

i'm sure i'm not the only one out there looking for this type of thing, yet i've found no such framework and no mention of anything like it.

thank you

[planetsim]

Is it possible yes its possible but you will need to forget to do the following tasks

Planning (Even a smallish site with very little interaction would need at least a day to plan properly)
Security - Of course you can add it in depending on how you do the initial coding and documenting of code you do
Design - well you can only do a mockup with basic layout mainly to show functionality

Even with a frame work this is still would be some trouble, I also dont know of any frameworks that would at least give you a startup to get going quickly as I dont use frameworks since I dont like any of them.

Now if your willing to do a half "assed" job to show for a week then go right ahead however remember that first impressions count so by all means this is really not possible, of course I dont know the complexity of the project so I cant really say more.

[justsomeone]

This is mad. I would expect that properly documenting and signing off on the requirements would take a few days at least, killing off half your week anyway. There's no way you'll build anything useful from scratch in the reamaing few days if you're doing all this on your own.

This doesn't sound like a client you need. My experience is that clients who insist on rush jobs don't get better as time progresses. They tend to refuse to sign off on requirements, withhold payment until you make loads of free changes (because you had to fill in the gaps in terms of guessing their requirements, and they didn't like what they saw.

If this client expects you to doa professional job, they need to do their job professionally as well.

Ditch this guy.

[Weedpacket]

RFC1925
(2) No matter how hard you push and no matter what the priority, you can't increase the speed of light.
(2a) (corollary). No matter how hard you try, you can't make a baby in much less than 9 months. Trying to speed this up might make it slower, but it won't make it happen any quicker.

Or if you want a more currently apropos way of putting it: "I cannae change the laws of physics, Captain!"

[hlslaughter]

it is a bit unrealistic to expect anything good in that short a time frame.

they're trying to break one of the most important laws of development, the old 'good, fast, cheap, pick any two'. they want all three.

I think they're either going to need to sacrifice speed or quality. I don't think you can demand a scalable, well thought out long-term solution and want something palpable in a week at the same time.

my thought was that if there were such a thing as a sane, thought out basic starter kit for a new website, i could get straight to development and hope that the framework would take care of the basics.

I do have a book called Secure PHP Development in which the author uses his own framework for building out 50 different web applications. My concerns with it are that it uses PearDB (which is supposed to introduce significantly more overhead than ADOdb). And I believe his templating system does not cache compiled templates (as does Smarty), so you're going to see the overhead of that element on each page as well.

You can view some of that framework here.

I am not sure what to do. I'd be uncomfortable hacking something together that quickly without putting any thought into it. There's another part of me that says I should just get to it and throw all common sense out the door 🙂

[hlslaughter]

A friend of mine suggested that TikiWiki might be what I'm looking for.

I have no experience with this, but my buddy claims that you can prototype your site using TW and then customize/optimize the elements you wish to, eventually replacing TW elements with your own code if you wish.

Anyone have any experience developing with TW?

[LoganK]

I've never used TikiWiki, but I do understand that a client that wants all that done in a week, especially taking into consideration normal "9-5" work days and contracts and all, is crazy. I wouldn't necesarily ditch him like some people have mentioned, but I'd let him know it'd take at least 2 weeks to get a basic framework complete, and around 4-6 weeks for a full site.

Edit: I mean if you really want to, you can go into "coding overdrive" and hold a red-eye coding session; i.e. coding that lasts for three days straight at a time, no sleep, no eat, no nothing. At least you'd get it done 😃

[rikmoncur]

Some clients seem to think that the world is gonna beat a path to their door as soon as the website goes online.
I've had clients frantic to get the site launched by Friday - and how many hits does their website get in the first month - um.... 'bout three?
They expect you to move into overdrive and do a rush job but how long does it take them to honour their invoice? No rush then!

[LoganK]

Good point 😃

[Weedpacket]

What I enjoy is when they put the URL on all their current advertising and then come around to get their site built, hosting arranged, domain name registered, and so on.

Y'know, like when you announce the location of your new superstore and the shopping public comes by to find an undeveloped section with a real estate agent's sign sticking out of it.

[LoganK]

I've encountered those types before...let me tell you they're the hardest to work with. Way to self-absorbed and impatient.

[planetsim]

LoganK wrote:

I wouldn't necesarily ditch him

I would, its not just about the client. Its about your image or your companies image also. If you willing to do a rush job are you going to add this site to your portfolio? If your not its not worth doing. If you are what are potential clients going to think when they see your portfolio, first impressions count like I said in my previous post. If they get an impression of a good site than check this there going to think your firstly inconsistent, unprofessional and whatever else they think. Although some clients come around if you educate them, majority wont and your likely to get some abuse if your doing this via IM or any other form that doesnt require face to face contact I guess its just the type of people they are "impatient and rude"

Again going back to my first post are the trade offs worth it? For me Ive learnt my lesson with my first and last client of this type. There truly not worth the effort no matter how much they throw at you (moolah that is).

[hlslaughter]

this isn't quite that bad. at least they aren't publishing the url yet 🙂

although it is about 6 non techies, one cto and me. so of course, everyone has grand ideas, big plans. and it's just me to build it! it's actually a cool opportunity for me if the cto can continue running interference. so far, he's pretty good at shutting down the marketing folks at the turn.

i'm leaning towards just doing everything from scratch by hand. if i can't meet their timelines, so be it. but i'm not going to panic about it or worry too much about meeting unrealistic goals. and i'm certainly not going to make the mistake i see so often made of brute forcing something together with code scraps found in a dirty alley somewhere.

but it does surprise me that with php now being 10 years old, nobody has put together a little tarball of a file structure with all the most basic components of every website in existence ready to go. all these CMS and framework projects seem to shoot for the sky.

[bubblenut]

I'd say stick with what PHP makes themselves if you don't have long to decide. There are some really cool frameworks out there but the time it would take to look through them all and decide which is right for your project you could have built it. When time is short you just need to get on and do it. For your DB abstraction layer I'd have to say PDO, it's the new big player on the DB abstraction scene and it's supposed to be pretty damned hot (I've not used it myself yet, just read up on it). Headed by Wez Furlong, one of the great minds in PHP at the moment, you can check out it's wiki page over here. The templating engine really depends on how big this site is, I really like Smarty but unless the site is going to be pretty big it's just overkill. By the sounds of things this site is going to be pretty small (or else you're in real trouble), in which case I'd probably suggest PEAR's Flexy it's small, simple but very powerfull and midbogglingly quick.

[Shrike]

A week should be enough to write the code. Agreeing the spec is what takes the time, and as you say when the spec isn't formalised properly the client has plenty of 'oh but...' space. However it's easy to say 'just ditch the client' when you have other work you can be doing, but something has to pay the bills 😉

Since you are on such a short timeframe I think forcing the client to pay half of the cost up front would be a good idea. Especially if you have no formal spec to work with.

I personally have a grudge against templating engines as a whole. Smarty is just a wrapper to PHP, it's compiled templates are PHP so why bother? To those who say 'but Smarty can be used by non-programmers': in my opinion PHP's syntax as much friendlier than Smarty's and by definition it is faster.

To set the record straight Smarty is a templating engine and not an MVC framework (look at Phrame, Mojavi etc). I have played with Phrame a bit but didn't like it much. Not tried Mojavi but I hear it's very good (possibly only PHP 5 now).

With regard to website components, most of what you want can probably be found between PEAR and www.hotscripts.com. But I think the reason why there is no definitive package for building webapps is that each have slightly different requirements and it's easier to tailor a suite of components which you've written yourself and hence know inside out.

[LoganK]

I've seen that term a lot too - MVC. What does it mean?

After reading your post planetsim, it might be better to ditch them. Although, sometimes if it's a big company, I might not ditch them because they might call on me in the future and I could be contracted for something else. It is all about the money, like you said 🙂

[bubblenut]

Oooh, I can't not reply to that. I agree, Smarty is too bloated for most sites but a distinct and seperated display layer can be very usefull. The use of a seperate syntax is questionable but the overall concept is very sound. No smarty is not an MVC in itself but it leds itself very nicely to being a V (ie the view component of an MVC). This is (pretty much) how things are handled where I am currently working and it works pretty well.
What I'm trying to get at is the fact that a templating engine is an important layer in any design. Smarty's decision to use a seperate syntax for their templates is questionable but it certainly has it's plus points. The access to php's functionality is restricted, threby preventing people who shouldn't be able to cock things up from doing so. The error handling is confined to the template so that designers only see designers errors. Just two points but I guess you can already see a pattern, something as feature rich and seperated as Smarty is only really of any use when you have a clear distinction between designers and developers. This is often not the case. It is, however, allways the case that a templating engine is usefull, even if this is just a very simple, assign variables and call up a php script with the display in kind of thing. Oh, another thing Smarty does which is pretty handy is language support so you can have your site in multiple languages.
I'm not quite so sure about pre-build frameworks. As yet I haven't used any but I'm not going to rule them out, a pretty good list of PHP frameworks can be found here. Prado looks pretty funky, hadn'y thought of using PHP quite such an event driven manner.

On a completely different subject, damn, how did I not find that wiki before today, it's awesome and going to be soooo cool when there's more stuff in it.

[Shrike]

I agree that separation of the view from the business logic is important, and not just for working in groups. Being able to have an HtmlView, an RssView and a WapView generated by the same logic has obvious benefits. I also agree that to achieve this there does need to be some method of passing data from the model to the view and this could be called a templating engine of sorts. I just think that the arguments for using Smarty don't bear upto scrutiny. The model itself should be abstracted away from the view, in fact the view shouldn't even know it exists. And if the designers start poking around they need to be hit with a LART.

Logan: MVC stands for Model View Controller, a 'classic' Gang of Four design pattern which has found a home in web development.

[LoganK]

Thanks for clearing that up!

[Jason_Batten]

For the right price, anything is possible.