rapidsai/asvdb

asvdb's Issues

unable to query a db on s3

If I query a db on s3 asvdb --read-from=s3://<bucket_name> --print="commitHash, argNameValuePairs, result, unit" I get the following error:

Traceback (most recent call last):
  File "/home/radek/miniconda3/envs/rapids-22.10/bin/asvdb", line 33, in <module>
    sys.exit(load_entry_point('asvdb', 'console_scripts', 'asvdb')())
  File "/home/radek/workspace/asvdb/asvdb/__main__.py", line 208, in main
    fromDb = openAsvdbAtPath(args.read_from)
  File "/home/radek/workspace/asvdb/asvdb/__main__.py", line 97, in openAsvdbAtPath
    db.updateConfFile()
  File "/home/radek/workspace/asvdb/asvdb/asvdb.py", line 233, in updateConfFile
    self.__updateConfFile()
  File "/home/radek/workspace/asvdb/asvdb/asvdb.py", line 514, in __updateConfFile
    raise AttributeError("repo must be set to non-None before "
AttributeError: repo must be set to non-None before writing /tmp/tmpmnj8z4ci/asv.conf.json

If I download the contents of the bucket and query the db locally, everything works.

I execute aws s3 sync s3://<bucket_name> <dir_name> and then asvdb --read-from=<dir_name> --print="commitHash, argNameValuePairs, result, unit"` and get the result:

[FEA] Support "union" of benchmark.json files

Is your feature request related to a problem? Please describe.
ASVDB should be able to take the results of multiple benchmarks.json files and create a "merged" benchmarks.json that contains all benchmark info.

Example Problem

1. Branch-0.14 stable release contains benchmark A, B, and C data
2. Branch-0.15 stable release contains benchmark B, and C data
3. Benchmark A data does not render in asv publish

benchmark.json will be based on most recent branch, causing us to not render the data of benchmark A. We should be able to merge these two benchmark jsons to make sure all data is represented.

Describe the solution you'd like
A new method that can parse a set of files. Best available on the command line with an argument to provide a filepath to stored .json files to merge.

Cannot open ASV results

Given a set of results generated by ASV.

run asvdb --read-from .asv/results

Traceback (most recent call last):
  File "/raid/kgerman/miniconda/envs/cuml_integrations/bin/asvdb", line 10, in <module>
    sys.exit(main())
  File "/raid/kgerman/miniconda/envs/cuml_integrations/lib/python3.7/site-packages/asvdb/__main__.py", line 210, in main
    results = fromDb.getResults()
  File "/raid/kgerman/miniconda/envs/cuml_integrations/lib/python3.7/site-packages/asvdb/asvdb.py", line 314, in getResults
    retList = self.__readResults(filterByInfoObjs=filterInfoObjList)
  File "/raid/kgerman/miniconda/envs/cuml_integrations/lib/python3.7/site-packages/asvdb/asvdb.py", line 430, in __readResults
    paramValues = benchmarkResults["params"]
TypeError: 'NoneType' object is not subscriptable