- If logfiles are not found when parsing then a FileNotFoundError is raised
- Parsing of heuristic solutions found prior to nodelog (but not found by NoRel) (#51)
- slightly reorganized parsers
- better parsing of early solutions (pre-tree and norel)
- rename project to gurobi-logtools for consistency with other Gurobi projects
- add Python 3.11 tests
- Command line option to copy/rename log files according to parameter settings (#34)
- Parsing of deterministic work metric in nodelog, barrier, and simplex (#27)
- Add a ChangedParams summary column with parameters as a dictionary (#21)
- Handle pandas warning related to groupby(); fix join issue in pandas2 (#29,#33)
- The code went through a refactoring while it remains backward compatible.
- Seed and Time columns in the DataFrame are reported as integer and datetime types, respectively
- The
get_dataframe(logfiles, timelines=False, prettyparams=False)
API has a new signature - A new API
parse(patterns)
is added that returns aParseResult()
object withsummary()
andprogress(section)
methods (see README.md for an example usage)
- fix error when log contains cuts with the name containing a hyphen
- fix error when log does not have a tree section
- refactor to make code more modular
- avoid coercing NewSolution to numeric type
- update default parameter list for v9.5.0
- add counter to LogFilePath to better distinguish merged logs
- new parameter
prettynames=True
to replace parameter values with meaningful descriptions
- fix parsing of MIP Start information
- add missing range attributes for quadratic constraints
- parse output from NoRel heuristic into separate DataFrame (included in
timelines
dictionary)
- new output format for timelines:
- second return value (when using option
timelines=True
) is now a dictionary including the individual timeline DataFrames, that isrootlp
,nodelog
, andnorel
- see README for updated usage information
- second return value (when using option
- support Compute Server log files
- add Cores count
- initial public release