Ruby on Rails is a web ap­pli­ca­tion frame­work, which is hyped to be the bomb. OK, the in­stal­la­tion was a breeze, as ad­ver­tized, but af­ter read­ing on a bit, it seems all so point­less to me. The out of the box Model-View-Con­troller pat­tern might steer a de­vel­op­er away from a meri­ad of spaghet­ti cgi scripts or php pages, since this is ex­act­ly what new­bies tend to very en­t­hou­si­as­ti­cal­ly screw up big time. It can be used to build some­thing that might be in­ter­pret­ed as a web­site in very lit­tle time with very lit­tle code. But then it prob­a­bly sud­den­ly gets com­pli­cat­ed. Ba­si­cal­ly all the frame­work does is map URLs to skinnable views avail­able in a web­brows­er. This is cool for Joe Scrip­tkid­die who want's to put some shit into a mySQL data­base and pro­gram a "user" in­ter­face around it, but over 10 years ago real data­bas­es could just do that by them­selves, and 5 years ago real data­bas­es of­fered a sim­i­lar MVC web-in­ter­face, and years ago oth­er MVC based ap­pli­ca­tions were avail­able. Why weren't they suc­cess­ful then?

If the Ruby on Rails ap­proach re­al­ly dec­i­mates the de­vel­op­ment time, it would be worth to in­vest in such an ap­proach. Maybe Or­a­cle or Sun had such an ap­proach, put a com­mer­cial pric­etag on it and off course no de­vel­op­er out of the 2-month-sin­gle-de­vel­op­er-project cat­e­go­ry was in­ter­est­ed in a prod­uct that ap­par­ent­ly saves 90k on a 100k project. And now these de­vel­op­ers try to push a "scal­able" so­lu­tion into the high-end?

The web is more than just a thin GUI to a data­base, and an MVC web ap­pli­ca­tion is not a Con­tent Man­age­ment Sys­tem. So, while I'm read­ing about Ruby on Rails I'm think­ing of the around 150 mul­ti-lin­gual con­tent man­aged web­sites with sev­er­al back-end con­nec­tions I'm in­volved in. Look­ing at the Ruby on Rails "suc­cess sto­ries" I won­der whether the de­vel­op­ers are blind, or it must be re­al­ly com­pli­cat­ed to use some im­ages to go with the text. Sure, from a data­base point of view it kind of screws up your MVC if you want an ar­bi­trary num­ber of im­ages to be able to be min­gled be­tween some text, and the back-end GUI to man­age them wouldn't be that straight for­ward any­more ei­ther. Which makes the clean frame­work pret­ty use­less be­yond the realm of self-in­dulged blog- and wiki sphere. It would be some­thing Mi­crosoft would get right be­cause they know their cus­tomers re­al­ly want nice im­ages...

Real­i­ty isn't al­ways clean and log­i­cal, and mod­els aim­ing for too much of it break eas­i­ly un­der the pres­sure of real de­mands. The client pro­vid­ing the busi­ness "log­ic" is just a hu­man and the vis­i­tor ex­pe­ri­enc­ing the UI is also just a hu­man, and in be­tween there is lit­tle need to go be­yond what they con­sid­er log­i­cal in their own - some­times stu­pid - sys­tem of ref­er­ence. Life is like a bunch of ever chang­ing hacks, and so­lu­tions should be able to han­dle them. Un­less you're on the bot­tom end im­ple­ment­ing Open Source so­lu­tions be­cause your clients can't even af­ford the pro­pri­etary al­ter­na­tive. But then you're prob­a­bly solv­ing an­oth­er prob­lem...

Another is­sue is in­creased beta sta­tus. For an Open Source project to be suc­cess­ful it needs de­vel­op­ers, prefer­ably be­fore the prod­uct is ready to use. But it's hard to pro­mote some­thing that doesn't work yet, so de­vel­op­ers have to be tricked into us­ing some­thing that isn't fin­ished yet on a pro­duc­tion en­vi­ron­ment, and have the lock in force them to fix the bugs. Ad­ding beta does just that. The only down­side to this is that the project starts to smell of un­sta­ble­ness which is hard to get rid off. Re­lease can­di­dates aren't re­al­ly re­leas­es, scal­a­bil­i­ty starts with sta­bil­i­ty. And I re­al­ly should stop read­ing slash­dot.