note cloc seems to be giving fairer results as it might be counting actual lines of code (ie, single lines of code taking multiple lines count as a single line of code, etc).
Gabriel Roldan wrote: > This may lay out better: > http://pastebin.com/m77cc8264 > > Gabriel Roldan wrote: >> There's also the count lines of code utility (cloc: >> http://cloc.sourceforge.net) >> >> This is the output for test and non test code: >> >> non test: >> $ cloc --exclude-dir=community,test . >> 3889 text files. >> 2591 unique files. >> 39287 files ignored. >> >> http://cloc.sourceforge.net v 1.08 T=50.0 s (42.1 files/s, 6458.2 lines/s) >> ------------------------------------------------------------------------------- >> Language files blank comment code scale 3rd >> gen. equiv >> ------------------------------------------------------------------------------- >> Java 1140 29912 71239 108562 x 1.36 = >> 147644.32 >> XSD 297 784 6694 47207 x 1.90 = >> 89693.30 >> XML 366 1352 2282 26054 x 1.90 = >> 49502.60 >> Javascript 87 902 1856 9353 x 1.48 = >> 13842.44 >> JSP 71 628 1277 4371 x 1.48 = >> 6469.08 >> HTML 123 671 174 4264 x 1.90 = >> 8101.60 >> CSS 13 336 193 2693 x 1.00 = >> 2693.00 >> DTD 4 283 732 308 x 1.90 = >> 585.20 >> PHP 1 51 67 290 x 3.50 = >> 1015.00 >> DOS Batch 2 72 17 160 x 0.63 = >> 100.80 >> Bourne Shell 2 14 21 89 x 3.81 = >> 339.09 >> ------------------------------------------------------------------------------- >> SUM: 2106 35005 84552 203351 x 1.57 = >> 319986.43 >> ------------------------------------------------------------------------------- >> >> test: >> $ cloc --exclude-dir=community,main . >> 2361 text files. >> 1698 unique files. >> 41930 files ignored. >> >> http://cloc.sourceforge.net v 1.08 T=33.0 s (37.2 files/s, 4267.4 lines/s) >> ------------------------------------------------------------------------------- >> Language files blank comment code scale 3rd >> gen. equiv >> ------------------------------------------------------------------------------- >> XSD 298 784 6698 48181 x 1.90 = >> 91543.90 >> XML 366 1272 2212 25991 x 1.90 = >> 49382.90 >> Java 248 4555 3470 18915 x 1.36 = >> 25724.40 >> Javascript 105 906 1910 9470 x 1.48 = >> 14015.60 >> JSP 71 628 1277 4371 x 1.48 = >> 6469.08 >> HTML 114 602 178 4036 x 1.90 = >> 7668.40 >> CSS 14 336 193 2696 x 1.00 = >> 2696.00 >> DTD 4 283 732 308 x 1.90 = >> 585.20 >> PHP 1 51 67 290 x 3.50 = >> 1015.00 >> DOS Batch 2 72 17 160 x 0.63 = >> 100.80 >> Bourne Shell 2 14 21 89 x 3.81 = >> 339.09 >> SQL 1 5 4 31 x 2.29 = >> 70.99 >> ------------------------------------------------------------------------------- >> SUM: 1226 9508 16779 114538 x 1.74 = >> 199611.36 >> ------------------------------------------------------------------------------- >> >> >> Andrea Aime wrote: >>> Andrea Aime ha scritto: >>>> Justin Deoliveira ha scritto: >>>>>> One thing that I was not sure about was the usage of the svn log to >>>>>> report numbers of test classes and the like. On one side, it's a way >>>>>> to give a concrete information about the testing, on the other side >>>>>> it might be perceived as trying to look (too) good. Opinions? >>>>>> >>>>> I think that information would be in general good to have, I would be >>>>> curious as to what the ratio of code to test code is. I guess whether we >>>>> are trying to "show off" or not would depend on the numbers... for >>>>> instance they may be embarrassing :) >>>> Rough but quick computation (on trunk, giving up on removing white >>>> space, comments and the like): >>>> cat `find . -name *.java` | wc -l --> 331976 >>>> cat `find . -name *.java` | wc -l --> 33885 >>>> non test code: 297991 >>>> % of test over non test: 11.3% >>>> >>>> Note this is a unfair towards test code, as non test code has >>>> javadocs, surely more code comments, and copyright headers. >>>> But it's a ballpark idea, I think it's fair to assume the actual >>>> ratio is more towards 15%-20% >>> Hmmm... making a better estimate is possible by removing whitespace >>> and comment lines. Here is an attempt trying to shave off white lines, >>> block comments and lines with just a comment (assuming for simplicity >>> that those starting with "*" are most of the time javadoc comments and >>> lines ending with */ are also the end of a multiline block comment): >>> >>> cat `find . -name *.java` | grep -v "^[ \t]*$" | grep -v "^[ \t]*//" | >>> grep -v "^[ \t]*/\*" | grep -v "^[ \t]*\*" | grep -v ".*\*/" | wc -l >>> 183556 >>> >>> cat `find . -name *Test.java` | grep -v "^[ \t]*$" | grep -v "^[ \t]*//" >>> | grep -v "^[ \t]*/\*" | grep -v "^[ \t]*\*" | grep -v ".*\*/" | wc -l >>> 24938 >>> >>> non test code: 158618 >>> % of test over non test: 13.58% >>> >>> Hmmm... I expected it to be higher, but it's not so bad. Most of our >>> tests are functional so they tend to exercise a lot of the GeoServer >>> code (startup, configuration, catalog access, data access, service code >>> and so on). >>> >>> Cheers >>> Andrea >>> >> > > -- Gabriel Roldan OpenGeo - http://opengeo.org Expert service straight from the developers. ------------------------------------------------------------------------------ _______________________________________________ Geoserver-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/geoserver-devel
