Implementation of distribution_t data structure #40
Conversation
src/daemon/distribution.c
Outdated
| }; | ||
| for (size_t i = 1; i < num_buckets; i++) { | ||
| bucket_array[i].bucket_counter = 0; | ||
| bucket_array[i].minimum = bucket_array[i - 1].maximum; |
There was a problem hiding this comment.
At your option: You could use a compound literal here, too:
bucket_array[i] = (bucket_t) {
.minimum = bucket_array[i - 1].maximum,
.maximum = (i == num_buckets-1) ? INFINITY : bucket_array[i - 1].maximum * factor,
};There was a problem hiding this comment.
I implemented such approach and realised that now my constructor functions are very similar. The main difference is in calculating maximum boundaries of buckets. Maybe, it would be better to select maximum boundaries first and then call the function that will fill other fields? Or such code duplication is okay?
…ubles equality check)
src/daemon/distribution_benchmark.c
Outdated
| } | ||
|
|
||
| int main() { | ||
| FILE *fout = fopen("benchmark.csv", "w"); |
There was a problem hiding this comment.
fopen could possibly not succeed, so I'd recommend to check if fout is equal to NULL and if yes, return.
For consideration:
There is also another possibility to redirect output to file. If you would use only printf, then running ./distribution_benchmark > benchmark.csv should return the same file.
There was a problem hiding this comment.
I agree with Barbara, for example I wrote to the csv file using printf and writing:
./distribution_benchmark $i >> results.csv
in the bash script.
src/daemon/distribution_benchmark.c
Outdated
| FILE *fout = fopen("benchmark.csv", "w"); | ||
| fprintf(fout, "Number of buckets,Average for update,Average for percentile,Total for %lu mixed iterations\n", MIXED); | ||
| for (size_t num_buckets = 20; num_buckets <= 4000; num_buckets += 20) { | ||
| distribution_t *dist = build(num_buckets); |
There was a problem hiding this comment.
I'd recommend to check if dist is NULL and do some things if yes
| static double *linear_upper_bounds(size_t num, double size) { | ||
| double *linear_upper_bounds = calloc(num, sizeof(*linear_upper_bounds)); | ||
| for (size_t i = 0; i + 1 < num; i++) | ||
| linear_upper_bounds[i] = (i + 1) * size; |
There was a problem hiding this comment.
This could possibly return an error because linear_update_bounds could be NULL. I'd recommend to check if linear_upper_bounds is NULL and do sth if yes
There was a problem hiding this comment.
Same in the other upper bounds functions.
src/daemon/distribution_benchmark.c
Outdated
| } | ||
|
|
||
| int main() { | ||
| FILE *fout = fopen("benchmark.csv", "w"); |
There was a problem hiding this comment.
I agree with Barbara, for example I wrote to the csv file using printf and writing:
./distribution_benchmark $i >> results.csv
in the bash script.
…-interns-2020 branch
Add distribution.h and distribution.c files that contains implementation of distribution_t data structure using the segment-tree approach.
You can read about the functionality of the distribution_t and the segment-tree approach in the design document: https://docs.google.com/document/d/1ccsg5ffUfqt9-mBDGTymRn8X-9Wk1CuGYeMlRxmxiok/edit?usp=sharing