I'd rather see too many underscores than abiglongvariablenameallasoneword. But then again, my text editor tries to autocomplete words (and does a fairly good job), so long variables aren't a big deal to me...

When I see people doing this kind of thing, I think "you must have a lot of time on your hands." Personally, I find that even after I use high-quality CPAN modules wherevere possible there is still plenty of coding to do, and the coding that's left is the part that is supplying actual value to my employer by solving their specific problem (as opposed to general problems like processing templates).

And what if there really is no template system that does what you need? Take one that's close and hack it, and submit the patches back to the maintainer. When Template Toolkit was not fast enough for us to use at eToys, we hacked it to make it much faster and then gave the code to the author. He used the ideas in it as the basis for a later release that we eventually ended up using.

A couple of times I have embedded extremely simple templating code into projects I have worked on (usually for code generation purposes). Is this wrong in general in your view? My own justification was that it was simpler to write the 15~20 lines of code to handle what I wanted than it was to use one of the pre-existing modules out there. Just getting aquainted with the API of the module would have taken longer than writing the code.

To turn this into a question, assuming you need a no-frills, quick, light and easy templating system what would you use? A simple

$tags=join "|",map {quotemeta $_} keys %tags;
1 while s/\b($tags)\b/$tags{$1}/g;
[download]

or a templating module?


---
demerphq

I would use Text::Template. It is very simple, fast enough, handles all the typical templating bugaboos (escaping, for example), and provides plenty of room for growth before you will bump your head and need a different module.