This patch implements a function called format_number_significant_figures() which is like format_number() but rounds to significant figures instead of decimal places. An accompanying patch for formatted_number module will soon follow.
I also wrote simpletests to test the significant figure rounding algorithm, in the attached file. I used these tests with simpletest 2.9 (note this version requires a core patch).
| Comment | File | Size | Author |
|---|---|---|---|
| #3 | format_number.test | 2.69 KB | Bevan |
| format_number.test | 2.81 KB | Bevan | |
| format_number_significant_figure.patch | 1.22 KB | Bevan |
Comments
Comment #1
Bevan commentedThe related patch for formatted_number module that uses the new function is at #580548: Additional formatters for fixed number of decimal places and significant figures.
Comment #2
Bevan commentedThe simpletest is a DrupalTestCase, but should really be a DrupalUnitTestCase. There is a bug in simpletest 2.9 that made that more problematic than it was worth though. That bug has now been fixed in Simpletest-dev and should be released in simpletest 2.10, whenever that comes out; #580406: DrupalUnitTestCase requires Database class
Comment #3
Bevan commentedThis is the simpletest DRUPAL-6--2 version of the test file that uses DrupalUnitTestCase.
Comment #4
markus_petrux commentedCould we add a link to a reference on how significant figures algorithm works? This would be fine as a note in the doxygen of the function itself. Then I will update the README and project page myself, but I would appreciate a link.
Comment #5
Bevan commentedAny suggestions on where to link to? I got the inspiration for the algo from here on php.net but it was broken and sub-optimal. I rewrote it myself. I contributed the corrected version back to php.net here, which is almost duplicate of what's in this patch.
Comment #6
markus_petrux commentedCommitted to CVS (commit). Thanks!
I have not committed the SimpleTest file. I'll add tests to the D7 branch, where the test framework is stable. For D6, it's been in flux, and I don't like the idea to keep supporting this yet.
Comment #7
Bevan commentedFair enough. Note that the signficant figures algo is sensitive so any changes should be tested with these simpletests, at the very least.
Comment #9
romano321 commentedThis is a calculator to use for this to check your code accuracy:
sig fig calculator
sig fig rules