add --total option. See https://github.com/brendangregg/FlameGraph/pu… #154
Conversation
The use case for this option is comparing multiple perf runs directly against one another. Specifying the # of samples to use for the total width means the individual rectangles in each graph can be directly compared to each other rather than being resized differently if one run finished more quickly than another.
|
This looks great, though it appears that it doesn't fully compile? |
The use case for this option is comparing multiple perf runs directly against one another. Specifying the # of samples to use for the total width means the individual rectangles in each graph can be directly compared to each other rather than being resized differently if one run finished more quickly than another.
Codecov Report
@@ Coverage Diff @@
## master #154 +/- ##
==========================================
- Coverage 88.4% 88.26% -0.14%
==========================================
Files 16 16
Lines 2200 2208 +8
==========================================
+ Hits 1945 1949 +4
- Misses 255 259 +4
Continue to review full report at Codecov.
|
|
Are there tests in the original PR too? If there are, would you mind porting them over? Otherwise, just copy an existing basic |
|
I'll take a look at adding a test, but it may be a few days before I have a chance. This is my first pull request on github, does everything here look ok? In particular, do I need to clean up the commits in my fork, or is ok to just have them all there separately like they are now? |
|
No rush — take whatever time you need :) That's exciting! Yes, it looks all good. You can tidy up your commit history if you wish, but otherwise I'll just squash them all into one commit before merging. It's probably not worth doing a lot of commit tweaking for a change this small, and individual commits are easier to review anyway! |
add --total option for generating the flame graph. see brendangregg/FlameGraph#14 and #17.
The use case for this option is comparing multiple perf runs directly against one another. Specifying the # of samples to use for the total width means the individual rectangles in each graph can be directly compared to each other rather than being resized differently if one run finished more quickly than another.