1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
|
# Comparing benchmarks for regression
The following commands are tested with `cabal` version 3.0.
Run the benchmarks for the baseline code i.e. without the changes:
```
# Remove any old benchmark results file first
$ rm results.csv
$ cabal run bench -- --csvraw=results.csv --quick
```
It will collect the benchmark results in `results.csv` file.
Now repeat the above step for the new code i.e. with the changes.
It will append the new benchmarks results to the existing `results.csv` file.
If you want more accurate benchmark results remove the `--quick` flag.
To generate a benchmark comparison between old and new changes from the
benchmark results in `results.csv` file:
```
# [NOTE] The path "results.csv" is harcoded in the chart executable.
$ cabal run chart --flag bench-show
```
# Comparing with ICU (text-icu package)
Install the icu library:
```
# On Mac OS using MacPorts
$ sudo port install icu
# On Mac OS using brew
$ brew install icu4c
# On Debian Linux based distributions (the library version suffix may differ)
$ sudo apt-get install libicu65
```
If cabal cannot automatically find the icu library (e.g. when installed
via macports install) then use explicit `LIBRARY_PATH` to tell it where
the library is:
```
$ export LIBRARY_PATH=/usr/lib/:/opt/local/lib
# Alternatively, pass the lib and include path as follows
$ cabal bench --extra-lib-dirs=/usr/local/opt/icu4c/lib --extra-include-dirs=/usr/local/opt/icu4c/include
```
Remove any old `results.csv` and run benchmarks with `has-icu` flag enabled:
```
$ rm results.csv
$ cabal run bench --flag has-icu -- --csvraw=results.csv --quick
```
The following command will now show the comparison between `text-icu` and
`unicode-transforms`:
```
$ cabal run chart --flag bench-show
```
|
