| Commit message (Collapse) | Author | Age |
|---|
| | |
|
| |
|
|
| |
As name was given tempfile object no name of generated temporally file.
|
| |
|
|
| |
This folder is no longer used and shouldn't be created.
|
| |
|
|
|
|
|
|
|
| |
Because configuration is now stored in database only missing
informations to make database fully descriptive is boot output. This
makes output of boot command to be saved to database and not to file.
Also parse script should read input from stdin and not from file from
argument.
|
| | |
|
| |
|
|
|
|
|
| |
Utils function callsubprocess now logs all program output including
stderr.
Convenient would be to print stderr even if output is disabled. But
this would probably required threads.
|
| |
|
|
|
| |
Previous implementation was returning list of tuples. Now it is
returning list of strings.
|
| |
|
|
| |
No attribute supplied when executed sql command.
|
| |
|
|
|
|
| |
Picosat output is parsed and automatically is resolved to dictionary
with configuration option name as key and boolean value. This makes all
configurations in script of same type. All are dictionaries this way.
|
| |
|
|
| |
This table was recently added, but was missing in databaseclean.sql.
|
| |
|
|
|
|
|
| |
This caused that instead of multiple numbers was parsed only one number
including every single number character as separated number. Because of
this was generated wrong CNF file and sometimes if zero was in number
picosat failed with error.
|
| |
|
|
|
| |
Temporally file generated in kernel script during configuration
applying wasn't closed/remove after end of usage.
|
| |
|
|
|
|
| |
Managing configurations in files and in database could cause
inconsistence. Adding all generated configurations to database allow us
to clean project files without loosing data.
|
| |\
| |
| |
| |
| |
| | |
Removing measuring branch. Idea was to make measuring configuration
separate from master. But keeping changes in master and measuring
branch consistent was pain.
|
| | |
| |
| |
| |
| |
| |
| | |
If generating single throw NoSolution exception, randomly generated CNF
file is removed. But finally statement tries to remove it once again.
Solution is to not remove CNF file in exception handling, but only by
finally statement.
|
| | |\
| | |
| | |
| | | |
Pull generate_single fix.
|
| | |\ \
| | | |
| | | |
| | | |
| | | |
| | | | |
Conflicts:
conf.py
scripts/configurations.py
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
Measure function requires img and config as atributes. But only config
was provided.
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
Configuration are generated with random assumptions. This should
generate different configuration for every SAT solver execution.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Previous method of calculating hash wasn't work (wouldn't give same
result for same configuration) if there would be added
new configuration options to hashconfigsort file.
|
| | | | | |
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | | |
As first will be measured only konfiguration options in kernel folder.
|
| | | | |
| | | |
| | | |
| | | | |
Loading configuration load all configuration options as True.
|
| | | | | |
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
When some generator tries to register already generated solution, it
shouldn't be registered.
|
| | | | |
| | | |
| | | |
| | | | |
Type inserted to method is str, but threated as int.
|
| | | | |
| | | |
| | | |
| | | | |
configuration per execution
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | | |
Parsing maximal value from first thread as result.
|
| | | | | |
|
| | |_|/
|/| |
| | |
| | |
| | | |
Executing parse_kconfig in test run makes no sense and takes to much
time.
|
| | |/
|/|
| |
| |
| | |
measure_list containst ints. But method writelines can only be used with
strings.
|
| | |
| |
| |
| |
| | |
This fixes and ensures that every call generates configuration. Unless
False is returned.
|
| | |
| |
| |
| |
| | |
Measure function requires img and config as atributes. But only config
was provided.
|
| | |
| |
| |
| |
| | |
Configuration are generated with random assumptions. This should
generate different configuration for every SAT solver execution.
|
| | | |
|
| | | |
|
| | |
| |
| |
| | |
Loading configuration load all configuration options as True.
|
| | | |
|
| | |
| |
| |
| |
| | |
When some generator tries to register already generated solution, it
shouldn't be registered.
|
| | |
| |
| |
| | |
Type inserted to method is str, but threated as int.
|
| | | |
|
| | |
| |
| |
| |
| |
| | |
Previous method of calculating hash wasn't work (wouldn't give same
result for same configuration) if there would be added
new configuration options to hashconfigsort file.
|
| |/
|
|
| |
configuration per execution
|
| |
|
|
|
| |
Creation of link to linux image wasn't done right.
Outcome was, wrong path in link.
|
| |
|
|
|
| |
Any exception raised from user script, that should parse boot command output, should be ignored.
If fails, no value is written to database.
|
| |
|
|
| |
This identifier is for recognition what generator generated specified configuration as first.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Also initial implementation of multithread execution.
A lot of functionality changed.
Phases removed.
Output parsing is now part of measure (boot) process.
Utils cleared.
Add dot_measure file for inverted dot_config.
Configuration generating is now prepared for multiple generating types. Fist implemented is generating configurations with single selected configuration.
Test is modified to be compatible with new changes.
|
