Remove watch_* operations in favor of "list_*(watch=True)" #9

mbohlool · 2016-11-17T01:17:54Z

Kubernetes supports two type of watch operations. One is the old deprecated one (watch_resource...) and on is the new one (list_resource...(watch=True)). k8sutil.Watch class support only the new one. It is not hard to support old deprecated ones too, but they are redundant and removing them make client less crowded. This PR removes old deprecated watch operations after it make sure there exists a new watch operation for them.

fixes #8

marun

Some style suggestions, the only one of note is the shadowing of id.

marun · 2016-11-19T00:40:43Z

scripts/preprocess_spec.py

+                                   make_operation_id_map)
+
+        def remove_watch_operations(op, parent):
+            id = op['operationId']


id is a builtin, so this variable assignment is effectively shadowing it. Consider differentiating via a trailing underscore (e.g. id_) or using a more verbose name (e.g. op_id).

marun · 2016-11-19T00:42:03Z

scripts/preprocess_spec.py

@@ -16,24 +16,63 @@ def _to_camel_case(s):
    return ''.join(_title(y) for y in s.split("_"))


-def iterate_through_operations(spec, func):
+def iterate_through_operations(spec, *funcs):


(No action required)

'iterate' usually implies an iterator, which this is not. Consider renaming to something like 'apply_funcs_to_spec'.

marun · 2016-11-19T00:43:14Z

scripts/preprocess_spec.py

@@ -16,24 +16,63 @@ def _to_camel_case(s):
    return ''.join(_title(y) for y in s.split("_"))


-def iterate_through_operations(spec, func):
+def iterate_through_operations(spec, *funcs):
    for k, v in spec['paths'].iteritems():
        for op in _ops:
            if op in v:


(No action required)

Consider avoiding unnecessary indentation:

for op in _ops: if not op in v: continue ...

marun · 2016-11-19T00:45:34Z

scripts/preprocess_spec.py

+        iterate_through_operations(spec, strip_tags_from_operation_id,
+                                   make_operation_id_map)
+
+        def remove_watch_operations(op, parent):


(No action required)

Consider avoiding the use of complex inline functions (longer than a few lines). They are effectively impossible to unit test because they can't be referenced outside of their scope.

Speaking of testing, consider adding tests for this script.

marun · 2016-11-19T00:47:09Z

scripts/preprocess_spec.py

+
+        def remove_watch_operations(op, parent):
+            id = op['operationId']
+            if not id.startswith("watch"):


(No action required)

Consider defining watch as a constant.

marun · 2016-11-19T00:59:13Z

scripts/preprocess_spec.py

+    for prop in prop_list:
+        if prop["name"] == property:
+            return True
+    return False


(No action required)

Consider being consistent in returning True/False or True/None (as with remove_watch_operations).

mbohlool · 2016-11-19T10:02:47Z

@marun Thanks for the review. I've applied your comments and also restructured the code to make it more readable. Please take another look.

marun

a couple of doc/naming nits, otherwise lgtm

marun · 2016-11-21T16:38:48Z

scripts/preprocess_spec.py

@@ -30,24 +39,67 @@ def _to_camel_case(s):
    return ''.join(_title(y) for y in s.split("_"))


-def iterate_through_operations(spec, func):
+def apply_funcs_to_spec(spec, func, *params):


nit: Given that this function only applies a single func, consider renaming to appy_func_to_spec

I missed that. Done.

marun · 2016-11-21T16:38:59Z

scripts/preprocess_spec.py

@@ -30,24 +39,67 @@ def _to_camel_case(s):
    return ''.join(_title(y) for y in s.split("_"))


-def iterate_through_operations(spec, func):
+def apply_funcs_to_spec(spec, func, *params):
+    """Apply fun to each operation in the spec.


nit: fun -> func

marun · 2016-11-21T16:39:03Z

scripts/preprocess_spec.py

+def apply_funcs_to_spec(spec, func, *params):
+    """Apply fun to each operation in the spec.
+
+    First two parameters of func will be operation and it's parent, and the


(No action required)

Consider defining the signature rather than explaining it (e.g. func(operation, parent)).

is it possible to define a function pointer's signature in python? you do you mean use that syntax in the comment?

Yes, use the syntax in the comment.

marun · 2016-11-21T16:39:17Z

scripts/preprocess_spec.py

+            return True
+
+
+def remove_watch_operations(op, parent, operation_ids):


(No action required)

If/when this method is targeted by a unit test, it may make sense to return messages (e.g. Cannot find...) as exceptions and allow the calling function more flexibility in how those cases are handled (i.e. script would output to stdout, test would check that the expected exception was returned for a given input).

marun · 2016-11-21T16:39:20Z

scripts/preprocess_spec.py

+def strip_tags_from_operation_id(operation, _):
+    operation_id = operation['operationId']
+    for t in operation['tags']:
+        operation_id = operation_id.replace(_to_camel_case(t), '')


(No action required)

to_camel_case is only referenced here, consider inlining.

Would like to keep this one :)

marun · 2016-11-21T16:39:24Z

scripts/preprocess_spec.py


-        iterate_through_operations(spec, strip_tags_from_operation_id)
+        operation_ids = {}
+        apply_funcs_to_spec(spec, make_operation_id_map, operation_ids)


(No action required)

make_operation_id_map is only referenced here, consider inlining. Given that this is a one-liner, it may be appropriate to use lambda.

marun · 2016-11-21T16:43:30Z

Is it necessary to include the generated commit in this PR (i.e. does the second commit fix things breakage in the generated code)? If not, consider breaking it the generated commit out into a separate PR that just gets merged.

mbohlool · 2016-11-21T18:24:50Z

In k8s, we always include the generated code commit in the same PR that changes the generating code. That will make sure the state of master after submitting each PR is valid (verify-* scripts will fail if we don't do that and we can't actually submit the PR). Here we do not have any verify script yet, but I still like to keep the master always in sync in regard to generating and generated code. I understand this is a suggestion not required. Can you elaborate more on how this is better? Thanks.

marun · 2016-11-21T18:31:52Z

Ok, makes sense. The code change results in a generated code change, I wasn't clear on that.

mbohlool · 2016-11-21T18:53:41Z

@marun I've made sure that changes to client here is related only to the changes in codegen in this PR by updating master generated code in #27

…*(watch=True)

mbohlool · 2016-11-21T21:52:50Z

@marun I think I applied all of your comments, please take another look and LGTM the PR if you are satisfied with the changes. Thanks.

marun · 2016-11-21T23:07:41Z

LGTM

Add tox and travis support

mbohlool assigned smarterclayton Nov 17, 2016

mbohlool force-pushed the o2 branch from 3e153a1 to c9d2d3d Compare November 17, 2016 01:21

mbohlool changed the title ~~Remove watch_* operations in favor of list_*(watch=True)~~ Remove watch_* operations in favor of "list_*(watch=True)" Nov 17, 2016

mbohlool assigned lavalamp Nov 17, 2016

mbohlool added this to the v1.0 milestone Nov 17, 2016

mbohlool added the kind/feature Categorizes issue or PR as related to a new feature. label Nov 17, 2016

mbohlool force-pushed the o2 branch 3 times, most recently from 75c10a7 to 2e2b12d Compare November 18, 2016 03:24

marun reviewed Nov 19, 2016

View reviewed changes

mbohlool force-pushed the o2 branch from 2e2b12d to 8a4a042 Compare November 19, 2016 10:03

marun reviewed Nov 21, 2016

View reviewed changes

k8s-ci-robot added the cncf-cla: yes Indicates the PR's author has signed the CNCF CLA. label Nov 21, 2016

mbohlool force-pushed the o2 branch from 8a4a042 to d31d286 Compare November 21, 2016 20:31

mbohlool added 2 commits November 21, 2016 13:24

Add preprocessing step to remove watch_* operations in favor of list_…

869fd7b

…*(watch=True)

Generated client

26b815f

mbohlool force-pushed the o2 branch from d31d286 to 26b815f Compare November 21, 2016 21:24

mbohlool assigned marun and unassigned smarterclayton Nov 21, 2016

mbohlool merged commit 880091d into master Nov 21, 2016

mbohlool deleted the o2 branch November 23, 2016 01:39

yliaog pushed a commit to yliaog/client-python that referenced this pull request Jan 8, 2022

Merge pull request kubernetes-client#9 from mbohlool/test

feee797

Add tox and travis support

		return True


		def remove_watch_operations(op, parent, operation_ids):

Remove watch_* operations in favor of "list_*(watch=True)" #9

Remove watch_* operations in favor of "list_*(watch=True)" #9

Uh oh!

Conversation

mbohlool commented Nov 17, 2016 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

marun left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mbohlool commented Nov 19, 2016

Uh oh!

marun left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

marun commented Nov 21, 2016

Uh oh!

mbohlool commented Nov 21, 2016

Uh oh!

marun commented Nov 21, 2016

Uh oh!

mbohlool commented Nov 21, 2016

Uh oh!

mbohlool commented Nov 21, 2016

Uh oh!

marun commented Nov 21, 2016

Uh oh!

Uh oh!

mbohlool commented Nov 17, 2016 •

edited

Loading