beat.backend.python issueshttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues2017-08-08T04:40:37Zhttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/10Improve test-ability of this package2017-08-08T04:40:37ZAndré AnjosImprove test-ability of this packageAs of today, it is not very easy to test the user process w/o an I/O daemon. We'd have to figure out a way to build unit tests around this package.As of today, it is not very easy to test the user process w/o an I/O daemon. We'd have to figure out a way to build unit tests around this package.André AnjosAndré Anjoshttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/9Remove prefix access and pass everything by value/json2017-08-08T04:40:37ZLaurent EL SHAFEYRemove prefix access and pass everything by value/jsonThe user process still has access to the prefix to fetch dataformats/algorithms/etc. Ideally, we need to get rid of this and to pass everything by value/json.The user process still has access to the prefix to fetch dataformats/algorithms/etc. Ideally, we need to get rid of this and to pass everything by value/json.André AnjosAndré Anjoshttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/8Split beat.core and beat.backend.python2017-08-08T04:40:37ZLaurent EL SHAFEYSplit beat.core and beat.backend.pythonFollowing a discussion with @andre.anjos , few beat.core functionalities need to be in beat.backend.python.
To achieve this, we need to:
1. Move all the relevant code from beat.core to beat.backend.python
2. Make beat.core depends o...Following a discussion with @andre.anjos , few beat.core functionalities need to be in beat.backend.python.
To achieve this, we need to:
1. Move all the relevant code from beat.core to beat.backend.python
2. Make beat.core depends on beat.backend.python
3. Make all packages use features from beat.backend.python (when they were moved from beat.core to beat.backend.python)André AnjosAndré Anjoshttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/7Do not open input/output pipe at each request2017-08-08T04:40:37ZLaurent EL SHAFEYDo not open input/output pipe at each requestAs discussed with @andre.anjos , this would be too inefficient.As discussed with @andre.anjos , this would be too inefficient.Laurent EL SHAFEYLaurent EL SHAFEYhttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/6Move algorithm and library into this package2017-08-08T04:40:37ZLaurent EL SHAFEYMove algorithm and library into this packageLaurent EL SHAFEYLaurent EL SHAFEYhttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/5Parameters handling requires a string to numpy-type conversion2017-08-08T04:40:37ZLaurent EL SHAFEYParameters handling requires a string to numpy-type conversionHandling the parameters of an algorithm requires the backend to be able to perform a string to numpy-type conversion.
This is currently performed by the method _convert_parameter_types() of the beat.core.algorithm.Algorithm class. Since...Handling the parameters of an algorithm requires the backend to be able to perform a string to numpy-type conversion.
This is currently performed by the method _convert_parameter_types() of the beat.core.algorithm.Algorithm class. Since the backend does not use this class, we need to move this function somewhere else to deal with experiment parameters.Laurent EL SHAFEYLaurent EL SHAFEYhttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/4Statistics split2017-08-08T04:40:37ZLaurent EL SHAFEYStatistics splitThe statistics are now generated by two different processes: the I/O daemon and the user process. However, everything is handled in one class, which means that statistics objects are often created but only half populated.
We may conside...The statistics are now generated by two different processes: the I/O daemon and the user process. However, everything is handled in one class, which means that statistics objects are often created but only half populated.
We may consider to split the class definition of statitics object to better reflect the underlying implementation.Laurent EL SHAFEYLaurent EL SHAFEYhttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/3Properly define IO daemon address2017-08-08T04:40:37ZLaurent EL SHAFEYProperly define IO daemon addressThe address of the IO daemon is currently hard-coded as localhost.
This should be properly defined and passed as an argument of the execute script.The address of the IO daemon is currently hard-coded as localhost.
This should be properly defined and passed as an argument of the execute script.Laurent EL SHAFEYLaurent EL SHAFEYhttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/2Move dataformat from beat.core2017-08-08T04:40:37ZLaurent EL SHAFEYMove dataformat from beat.coreIdeally, we don't want this package to depend on `beat.core`. But it requires dataformat.
One possible solution would be to move the dataformat from beat.core into this package, and to make beat.core depeonds on this package.Ideally, we don't want this package to depend on `beat.core`. But it requires dataformat.
One possible solution would be to move the dataformat from beat.core into this package, and to make beat.core depeonds on this package.Laurent EL SHAFEYLaurent EL SHAFEYhttps://gitlab.idiap.ch/beat/beat.backend.python/-/issues/1IO synchronization is managed twice2017-08-08T04:40:37ZLaurent EL SHAFEYIO synchronization is managed twiceThe I/O synchronization is currently manage twice: in `beat.io` and in `beat.backend.python`.
Since `beat.io` is common to any backend, it would be nice to only manage I/O synchronization in this package to keep the backends as simple...The I/O synchronization is currently manage twice: in `beat.io` and in `beat.backend.python`.
Since `beat.io` is common to any backend, it would be nice to only manage I/O synchronization in this package to keep the backends as simple as possible.Laurent EL SHAFEYLaurent EL SHAFEY