Refactor Attribute/Module/Pyrpl (not ready at all) #83

SamuelDeleglise · 2016-11-14T21:00:16Z

This is not ready at all:

no unit testing done yet
never even tried on python 2
still needs a lot of new tests and code documentation

However, it is functional for some stuffs (on python 3) and the main structure is not expected to move a lot now. In particular, NetworkAnalyzer is the first high-level Module to be up-and-running. I am very interested by some feedback (once your schedule becomes less crazy)

Main changes:

Attributes:

The basic class for a field that can be get/set graphically is an Attribute (see attribute.py)
Obviously the Registers also inherit from Attribute (inside attributes.py)

Modules:

Modules are now split in HardwareModules mostly made of Registers (formerly redpitaya_modules) and SoftwareModules mostly made of basic Attributes. The aim is to make Lockbox only one SoftwareModule amongst others (NetworkAnalyzer, SpectrumAnalyzer, CurveFitter).
HardwareModules are only available to the user via some ModuleManagers (IqManager, AsgManager, PidManager, ScopeManager), that deal with reserving/freeing the module.

Gui:

Only 3 files that map 1-to-1 the code structure:
attribute_widgets.py, module_widgets.py, pyrpl_widget.py
Still possible to completely disable the gui via a config file option
Gui is not updated by timers anymore, but by the set function of BaseAttribute.
Slave module (reserved through ModuleManager) have their widget disabled (+ title changed to show who owns them)
ModuleWidgets can now be closed and open again via a menu of the pyrpl_widget

Config file:

Each module instance has one top-level-entry in the config file (except for modules that have been slaved forever). This means the configuration of the lockbox signals, lock sequence... and so on will be done inside the lockbox section.
2 top-level items are an exception: repitaya (ip adress, login, ...) and pyrpl (gui or not, log_level, window positions, modules to load) on this last point, users only interested in low-level modules can simply remove Lockbox and CurveFitter from this list. Also, a possibility would be to use the same option in case only the curve fitter is needed (without hardware available).
Attributes that are in the list module.setup_attributes are saved automagically in config files with the same technic as for gui update. For instance, at start-up, the na is in the same state it was left before. In case the user wants to start with a fresh configuration, he can simply delete the corresponding top-level entry in the config-file.
It's always possible to prevent some module classes from saving their state by not instantiating the setup_attribute list.

lneuhaus · 2016-11-30T11:32:04Z

The entire discussion about this pullrequest is in issue #84 (should rather have been here but now it is a little late)

lneuhaus · 2016-11-30T11:38:46Z

pyrpl/attributes.py

+# docstring does not work yet, see:
+# http://stackoverflow.com/questions/37255109/python-docstring-for-descriptors
+# for now there is a workaround: call Module.help(register)
+class BaseRegister(object):


Sure we dont want from_python and to_python from intregister here?

lneuhaus · 2016-11-30T11:39:55Z

pyrpl/attributes.py

+        return obj._read(addr)
+
+
+class NumberRegister(BaseRegister):


Why is the class NumberRegister even necessary? The constructor that does not make use of the bits argument may be quite misleading, or is there a reason?

lneuhaus · 2016-11-30T11:42:28Z

pyrpl/attributes.py

+class FrequencyRegister(FloatRegister, FrequencyAttribute):
+    """Registers that contain a frequency as a float in units of Hz"""
+    # attention: no bitmask can be defined for frequencyregisters
+    CLOCK_FREQUENCY = 125e6


this is a typical case where we might want to move the definition to a global_config.yml file, such that the code allows a clean migration to other hardware platforms

lneuhaus · 2016-11-30T11:48:14Z

pyrpl/attributes.py

+    """
+    widget_class = FrequencyAttributeWidget
+
+    def __init__(self, default=None, increment=0.1, min=0, max=125e6/2, doc=""):


again, the clock frequency is quite hidden here... -> move to config file?

especially, because the actual clock frequency is 125e6*frequency_correction

lneuhaus · 2016-11-30T11:54:06Z

pyrpl/attributes.py

+
+    def __init__(self, default=None, increment=0.001, min=-.1, max=1., doc=""):
+        super(FloatAttribute, self).__init__(default=default, doc=doc)
+        super(FloatAttribute, self).__init__(default=default, doc=doc)


is this doubling desired?

SamuelDeleglise · 2017-01-27T17:53:20Z

Oh yeah baby !!!!

Not 100% sure though what makes them sometimes fail when they are executed all together, depending on the order...

lneuhaus · 2017-01-27T18:46:39Z

Python 2.7 isnt on your side:

======================================================================
ERROR: pyrpl.test.test_attribute_widgets.TestClass.test_dummy_module

Traceback (most recent call last):
File "C:\Python27\lib\site-packages\nose\case.py", line 197, in runTest
self.test(*self.arg)
File "W:\pyrpl\pyrpl\test\test_attribute_widgets.py", line 31, in test_dummy_module
assert(isinstance(self.pyrpl.dummy_module.true_or_false, bool))
AttributeError: 'Pyrpl' object has no attribute 'dummy_module'

======================================================================
ERROR: pyrpl.test.test_hardware_modules.test_scope.TestClass.test_scope_rolling_mode_and_running_sta
te_update

Traceback (most recent call last):
File "C:\Python27\lib\site-packages\nose\case.py", line 197, in runTest
self.test(*self.arg)
File "W:\pyrpl\pyrpl\test\test_hardware_modules\test_scope.py", line 29, in test_scope_rolling_mod
e_and_running_state_update
self.r.asg1.frequency = 0
File "W:\pyrpl\pyrpl\modules.py", line 384, in setattr
super(BaseModule, self).setattr(name, value)
File "W:\pyrpl\pyrpl\attributes.py", line 62, in set
self.value_updated(instance, value) # lauch signal and update
File "W:\pyrpl\pyrpl\attributes.py", line 88, in value_updated
self.save_attribute(module, value)
File "W:\pyrpl\pyrpl\attributes.py", line 118, in save_attribute
module.c[self.name] = value
File "W:\pyrpl\pyrpl\modules.py", line 301, in c
manager_section = getattr(self.parent.c, manager_section_name)
File "W:\pyrpl\pyrpl\memory.py", line 198, in getattribute
if isbranch(self[name]):
File "W:\pyrpl\pyrpl\memory.py", line 217, in getitem
return self._data[item]
File "C:\Python27\lib\site-packages\ruamel\yaml\comments.py", line 426, in getitem
return ordereddict.getitem(self, key)
KeyError: 'asgs'

======================================================================
FAIL: pyrpl.test.test_attribute_widgets.TestClass.test_config_file

Traceback (most recent call last):
File "C:\Python27\lib\site-packages\nose\case.py", line 197, in runTest
self.test(*self.arg)
File "W:\pyrpl\pyrpl\test\test_attribute_widgets.py", line 28, in test_config_file
assert("DummyModule" in self.pyrpl.c.pyrpl.modules)
AssertionError:

lneuhaus · 2017-01-27T18:48:20Z

FAIL: pyrpl.test.test_lockbox.TestClass.test_delete_output

Traceback (most recent call last):
File "C:\Python27\lib\site-packages\nose\case.py", line 197, in runTest
self.test(*self.arg)
File "W:\pyrpl\pyrpl\test\test_lockbox.py", line 45, in test_delete_output
assert len(widget.all_sig_widget.output_widgets)==old_len-1
AssertionError:

lneuhaus · 2017-02-04T22:40:45Z

pyrpl/test/test_ownership.py

+            if isinstance(module, ModuleManager):
+                with module.pop("foo") as mod:
+                    assert(mod.owner=='foo')
+                assert(mod.owner==None)


should be is

…t is not a module, i.e. only for top-level modules

…intain

….com/lneuhaus/pyrpl into refactor_register_submoduleattribute

dict keys/values are not indexable in python 3 (makes sense because there is no ordering)

dict keys/values are not directly indexable

…com/lneuhaus/pyrpl into refactor_registers

…grator is involved)

…into refactor_registers

…tor_registers

lneuhaus · 2017-04-02T23:52:58Z

pulling this into master now because it passes tests on travis and is more stable than original master version (release 0.9.1)

lneuhaus reviewed Nov 30, 2016

View reviewed changes

lneuhaus mentioned this pull request Nov 30, 2016

Point on refactoring #84

Closed

This was referenced Jan 16, 2017

Renaming of Variables #40

Closed

module.params should be uniformized #65

Closed

lneuhaus reviewed Feb 4, 2017

View reviewed changes

leo10tt10 and others added 17 commits March 8, 2017 02:08

fixes in memory and lockbox gui polished

6a3d7f0

state functional, load_setup_attributes call delegated to pyrpl"

68d18bf

slight change of stategy: module calls load_setup_attributes if paren…

8894477

…t is not a module, i.e. only for top-level modules

na starts to work with new API

7fe2ef6

defaultbranches removed from memory.py, was never used and hard to ma…

be84b7d

…intain

module_property instantiates itself upon first call of getter

338838e

stages are properly saved

15ef7ea

NA seems to work, no unittest yet

4a72c7d

Merge branch 'refactor_register_submoduleattribute' of https://github…

187777f

….com/lneuhaus/pyrpl into refactor_register_submoduleattribute

stages are visible in gui

d3ad70a

sequence doesnt work any more. Why?

db3aaa4

merged and resolved conflicts

7546717

NA,SA, Scope unitests are OK with the new "run" API

357acc2

Small changes to make Lockbox python 3-compatible

4fa6092

dict keys/values are not indexable in python 3 (makes sense because there is no ordering)

Small changes to make Lockbox python 3 compatible

b758194

dict keys/values are not directly indexable

adds Benchmark async sleep figures

bcec3c5

changes in timer benchmark

3f97182

leo10tt10 and others added 28 commits April 2, 2017 13:40

na running states bug on travis fixed

bfb2bd3

test_read_write_time still fails on travis

3ad9c17

comm time

efd247e

rw time

73d4d67

trying to fix all tests

7cb9f43

merge with resolved conflicts from refactor_registers_working_280317

47bf32b

non-scientific notation in global_config

2ecd235

fixes nosetests error with output

7d4beb8

Merge branch 'refactor_registers_bitfile_compiled' of https://github.…

aabefe9

…com/lneuhaus/pyrpl into refactor_registers

finetuning travis

84d247e

minor bugs

4b1d2d3

travis relaunch

c7ab29a

drastically reduced .travis.yml

b3e70c0

travis fix

20288e7

corrections

8e15fed

test_pid_na_2 is slightly affected by long delay (okay, since an inte…

3ee2670

…grator is involved)

exclude py3.6

f792d18

test python 3.5 before 2.7

f050bc9

delete curves after scope test

37439e3

fixes na_test issue for good

c0e1159

exclude python 3.3 from travis

3e35345

Merge branch 'refactor_registers' of https://github.com/lneuhaus/pyrpl …

858d318

…into refactor_registers

quamash for python >= 3.4

3742d8e

global_config_fix

a9e94c4

another config bug

4d5c300

travis order

267c0dd

curve delete bug removed

19bc7a2

Merge branch 'master' of https://github.com/lneuhaus/pyrpl into refac…

84fdb7c

…tor_registers

lneuhaus merged commit 012ae95 into master Apr 2, 2017

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Refactor Attribute/Module/Pyrpl (not ready at all) #83

Refactor Attribute/Module/Pyrpl (not ready at all) #83

SamuelDeleglise commented Nov 14, 2016

lneuhaus commented Nov 30, 2016

lneuhaus Nov 30, 2016

lneuhaus Nov 30, 2016

lneuhaus Nov 30, 2016

lneuhaus Nov 30, 2016

lneuhaus Nov 30, 2016 •

edited

Loading

lneuhaus Nov 30, 2016

SamuelDeleglise commented Jan 27, 2017 •

edited

Loading

lneuhaus commented Jan 27, 2017

lneuhaus commented Jan 27, 2017

lneuhaus Feb 4, 2017

lneuhaus commented Apr 2, 2017

Refactor Attribute/Module/Pyrpl (not ready at all) #83

Refactor Attribute/Module/Pyrpl (not ready at all) #83

Conversation

SamuelDeleglise commented Nov 14, 2016

lneuhaus commented Nov 30, 2016

lneuhaus Nov 30, 2016

Choose a reason for hiding this comment

lneuhaus Nov 30, 2016

Choose a reason for hiding this comment

lneuhaus Nov 30, 2016

Choose a reason for hiding this comment

lneuhaus Nov 30, 2016

Choose a reason for hiding this comment

lneuhaus Nov 30, 2016 • edited Loading

Choose a reason for hiding this comment

lneuhaus Nov 30, 2016

Choose a reason for hiding this comment

SamuelDeleglise commented Jan 27, 2017 • edited Loading

lneuhaus commented Jan 27, 2017

====================================================================== ERROR: pyrpl.test.test_attribute_widgets.TestClass.test_dummy_module

====================================================================== ERROR: pyrpl.test.test_hardware_modules.test_scope.TestClass.test_scope_rolling_mode_and_running_sta te_update

====================================================================== FAIL: pyrpl.test.test_attribute_widgets.TestClass.test_config_file

lneuhaus commented Jan 27, 2017

FAIL: pyrpl.test.test_lockbox.TestClass.test_delete_output

lneuhaus Feb 4, 2017

Choose a reason for hiding this comment

lneuhaus commented Apr 2, 2017

lneuhaus Nov 30, 2016 •

edited

Loading

SamuelDeleglise commented Jan 27, 2017 •

edited

Loading

======================================================================
ERROR: pyrpl.test.test_attribute_widgets.TestClass.test_dummy_module

======================================================================
ERROR: pyrpl.test.test_hardware_modules.test_scope.TestClass.test_scope_rolling_mode_and_running_sta
te_update

======================================================================
FAIL: pyrpl.test.test_attribute_widgets.TestClass.test_config_file