On 2013-02-16 02:43, Christian Hammond wrote:
On Feb 15, 2013, at 1:26 PM, Matthew Woehlke <mwoehlke.fl...@gmail.com> wrote:
Drifting even further off topic, have you ever given thought to using patience 
diff in RB? (I've seen spots where it would have been an improvement... might 
help with better moved-lines detection also. For that matter, what would be 
REALLY cool would be to compute common lines across the entire patch, to get 
cross-file line move detection :-).)

We haven't looked into patience diff. It might be worth looking into, but I 
know my TODO list is too long to consider rewriting the diff generator with a 
new algorithm :)

In my experience with git, patience diff generates much more reasonable diffs in many cases. I would strongly encourage you to at least do a web search and read one or more articles on the topic.

I wonder if you aren't already doing part of the work (finding common lines) for line move detection... (or if the way patience does line matching might be more efficient that what line move detection currently does...)

We want to be able to extend move detection across files. [...]
The reason it's not something we'll just be able to add today is that our diff 
generation is done on a per-file basis, one-by-one. You see this when loading 
the diff viewer. We need to lazy-generate because otherwise, large diffs would 
cause load times so long that they'd time out. We'd need to offload the diff 
scanning and generation to a background service, which requires more 
infrastructure than we'd like to require for a standard Review Board install.

For moved-line detection, do you at least scan the diff itself and not the patched files? (Since unmodified lines obviously will not have moved, this should be reasonably efficient except on reasonable sized diffs.)

For that matter, if performance is a problem, it may be a good idea to turn off more expensive features for very large diffs.

--
Matthew

--
Want to help the Review Board project? Donate today at 
http://www.reviewboard.org/donate/
Happy user? Let us know at http://www.reviewboard.org/users/
-~----------~----~----~----~------~----~------~--~---
To unsubscribe from this group, send email to 
reviewboard+unsubscr...@googlegroups.com
For more options, visit this group at 
http://groups.google.com/group/reviewboard?hl=en
--- You received this message because you are subscribed to the Google Groups "reviewboard" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to reviewboard+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.


Reply via email to