Suggested performance enhancements about daisydiff

[GoogleCodeExporter]

As with all other patches, it would be good if this was peer reviewed by at 
least one other person. I have some diff tests around that I can try, but it 
would be great if people applied this patch locally and run it against their 
own datasets.

Original comment by [email protected] on 6 May 2011 at 8:02

• Changed state: Accepted

[GoogleCodeExporter]

According to my 1000+ test cases this patch

A. Does not break anything.
B. Offers no noticeable speed regressions.

My tests run about 1-2% percent faster which is close to the statistical error.

Has anybody else run any benchmarks? I will perform some more detailed tests 
but unless there is an objection, I think this can go safely to trunk.

Original comment by [email protected] on 10 May 2011 at 7:15

• Changed state: Started
• Added labels: Type-Enhancement
• Removed labels: Type-Defect

[GoogleCodeExporter]

Hmmm... I just noticed that all my tests use the DaisyDiff "tag" mode, while 
the patch is against the "html" mode. This is why the behavior is essentially 
unchanged.

I will see if I can convert them to "html" mode and run them again.

Original comment by [email protected] on 10 May 2011 at 7:31

[GoogleCodeExporter]

The fixes are definitely specific to html mode.  They also may make little 
difference to documents with very shallow trees.

Original comment by [email protected] on 16 May 2011 at 4:40

[GoogleCodeExporter]

I don't seem to have big enough tests. Those that I have tried have 5%-20% 
speedup. They are mostly html tables.

Anyway I think this can go to trunk. Anybody has a different view on this patch?

Original comment by [email protected] on 19 May 2011 at 8:30

[GoogleCodeExporter]

Applied in r150

Original comment by [email protected] on 28 May 2011 at 6:57

• Changed state: Fixed