Sunday 26 May 2013

FOCUS! Oooh, shiny, shiny...

That didn't last long; having a nice set of components to build on and being satisfied that it would carry us forward indefinitely…

As Ruby developers on Rails or Padrino or bare metal, we're absolutely spoilt for choice. As I write this, Rubygems claims 1,577,138,099 downloads of 56,183 gems since July 2009. (Compare to ~495 PHP packages in the official PEAR repo on Github.) If you can figure out what you're looking for well enough for a Google search, chances are excellent that you'll find something you can at least use; likely several somethings. As with all such, this is both a blessing and a curse.

The blessing is that the stuff is out there; it is mostly at least usable, especially for the gems that are at or past Version 1.0 (yes, there's a formal numbering system for gems); and if it doesn't do precisely what you want, chances are likewise excellent that you can wrap your own code around it to do your bidding, and/or pull a copy of the official source, make changes you'd like to see, and submit a pull request to tell the original developer(s) what you'd like to see pulled into the "official" version. ANd so on.

Back a couple of years ago when I was doing PHP for a living, it wasn't that hard to keep a rough idea of what was both available and usable on PEAR in my head, so that as I developed a project, I knew what I had available and could try to structure my code to work effectively with it. It's a lot easier when you can reverse that: think of what you want and how you'd like to work with it, and go run down the list of candidates to plug that hole.

But that touches on the other edge of the sword, and it's sharp. Doing all that browsing and cataloguing and selecting and integrating takes time and effort; time and effort that your customers (external or otherwise) generally assume you're putting into making a kick-ass product. Well, you are, sort of; you're just trying to figure out which giants' shoulders to stand on. Throw in the usual social media like Google+ and Twitter and so on, alerting you to the newest and shiniest of the new-and-shiny, and it's a wonder you get any "real work" done.

Memo to bosses: This is real work, and it's almost certainly less expensive real work than having your folks reinvent several dozen (or several hundred) wheels as they develop your Awesome Product, Version next.0. Deal.

But the "is it done yet?" bosses do have a very good point, devs. Promises have been made; contracts have been let; people's jobs (including yours) may well be on the line. Focus is necessary. You just have to learn to do it in a reasonably structured way, like your RSI breaks.

When you come across the latest shiny-shiny that you just know is going to make your job so much easier, better, more fun and profitable, add it to a list. When you're about to start on some new part of your project, glance at the list and see if it makes sense to use your new feature-thingy as a test bed for the new gem. Be careful that it can peacefully coexist with the rest of your codebase that hasn't been updated yet, and do remember to use a new feature branch. Set yourself a time limit for how long you'll allow yourself to get the new gem working. When that's expired, give yourself at most one extension (hey, you're exploring the not-yet-known-to-you, after all) and if it's still not walking and chewing gum at the same time, go back to where you branched and try again, the "old" way. This is so simple, yet not doing it can have a devastating effect on your schedule (and project, and…)

But given a choice between running the risk of distraction by a myriad assortment of Things That Generally Work, and trying to make one of a smaller number of Things That Should Work But Don't, Quite… well, that's what brought so many VB6 and Java folk to PHP in the first place, and what brought so many PHP people (including yours truly) into Ruby. Given sufficient resources and patience, you can make almost anything work in almost any language; I once saw a quite-playable version of the classic Star Trek game written in RPG II. I thought, and still think, that the guy who wrote it was absolutely nuts (especially since he knew at least three other languages that would have been easier/better aligned to the job), but *it worked*. So, as the adage goes, never argue with success. However, given the chance early enough in the project for it to make a difference, do respectfully suggest that jobs are best done with the available tools most closely aligned to the job. RPG is still in production for what it does best: complicated report generation from highly-structured data. PHP remains a valid choice for small projects, or for those companies unwilling to rely on non-local talent and willing to trade initial cost for quality. Ruby has numerous tools available, for just about anything, that challenge its practitioners to be better developers than they started out to be. Part of that challenge is the sheer wealth of tools available, and their haphazard discoverability. Part of it is looking back at code you wrote a few short months, or even weeks, ago; knowing it could be so measurably better; and having the self-discipline to leave it be until you have a business reason to change it.

Problems like that make my days more enjoyable than solutions in other languages generally have. Now if I could just turn this megalithic Rails app into something more componentised, service-supporting. Oooh, shiny, shiny…