C. wrote:
PHPPeanuts (automates development but can be tricky to customize)
Prado (XML based meta language system)
Seagull (library type system with sensible functionality)
Zend (likely to be favoured in commercial web 2.0 type apps)
Blueshoes (lots of javascript eye candy)
CakePHP (very OO)
Horde (PEAR based)
Very nice, these characterizations. I am planning to evaluate my own
framework (phpPeanuts) against other frameworks, this is a nice
shortlist to start with. Thanks!
PHPPeanuts (automates development but can be tricky to customize)
The automation is most effective with applications that mainly store and
retrieve data (including searching). Currently i count with a
productivity between 9 and 22 function points per hour* depending on the
amount of custom tasks they require.
I agree customization does take quite some knowledge of the framework
and browsing through its code, but OTOH, you can still resort to
'manual' development of the parts that are too hard to customize or
simply not available.
However, the assumptions the framework makes about naming conventions
with respect to keys and foreing keys of your database tables are not
customizable, if you can not follow them it won't work.
Greetings,
Henk Verhoeven,
www.phpPeanuts.org.
* I currently estimate this kind of development as follows:
hours per
-----------------------------------
1 datatype (not counting pure n-m relation tables)
0,5 1-n relation
1 n-m relation
0,25 data field
2 custom data entry layout
0 simple search option
0 page/window
0 standard report
Then i add time for non-standard development tasks, for communication
with the customer, for integration testing and for delivery/installation.
In order to measure my productivity i relate this to function points
counting as follows:
points per
----------------------------------
8 datatype
4 1-n relation
8 n-m relation
4 data field
1 custom data entry layout
2 simple search option
4 page/window
6 standard report
For example phase two of the application i am currently building
consists of 14 datatypes, 11 1-n relations, 6 n-m relations, 44 data
fields, 1 custom data entry layout, 44 simple search options, 44 pages,
42 reports and custom task. It is estimated to take 40 hours. This
application phase counts for 896 function point, so i am doing 22.4
function points per hour.
Now of course the more custom tasks i have the lower is productivity:
the third phase is estimated at 270 points in 30 hours, which is only 9
fp/hr.
(These function points are based on measured productivity when i started
automating development. I define productivity by then as 1 function
point per hour) (we are talking 1996, GUI based applications using an
IDE with a screen painter, a O-R mapping class library and event-based
programming in Visual Smalltalk).
