The best architecture I saw out there (look at this guys old YAPC presentation) put the raw report data into XML form and used XSLT on that to create HTML, CSV, PDF, Excel, etc. To my mind this seems to be the lowest effort way of doing it, because with any luck, you just need a single piece of XSLT for each one of the output file types with a second piece to customize colors, fonts, etc.
As far as a report description format goes, XML isn't particularly human readable IMO, so I'd advise _not_ using XML. With a flat text file report description, you can create new reports the Unix Way (TM): with a text editor.
I'm working on my own report generator package in between all my pressing emergencies.
Until that's finished, all the stuff I need to put into production tomorrow I do with TT using plugins, which gives me the most report for the least programming.
In reply to Re: Avoiding reinventing the wheel: report generation systems?
by cleverett
in thread Avoiding reinventing the wheel: report generation systems?
by rkg
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |