Re: pgAdmin 4 commit: Support EXPLAIN on Greenplum. Fixes #3097

Поиск
Список
Период
Сортировка
От Dave Page
Тема Re: pgAdmin 4 commit: Support EXPLAIN on Greenplum. Fixes #3097
Дата
Msg-id CA+OCxoxC4Y7oYDM82XVuY8d-8YCJ_8tNu=CYjCj8KhXFrOdJgw@mail.gmail.com
обсуждение исходный текст
Ответ на Re: pgAdmin 4 commit: Support EXPLAIN on Greenplum. Fixes #3097  (Joao De Almeida Pereira <jdealmeidapereira@pivotal.io>)
Ответы Re: pgAdmin 4 commit: Support EXPLAIN on Greenplum. Fixes #3097  (Joao De Almeida Pereira <jdealmeidapereira@pivotal.io>)
Список pgadmin-hackers
Hi

On Fri, Mar 9, 2018 at 3:54 PM, Joao De Almeida Pereira <jdealmeidapereira@pivotal.io> wrote:
Hello,
Definitely running single tests is something that would be great, specially if you are TDDing something waiting 30-40 seconds to get feedback is a little cumbersome when the test you are concerned with take less then a second.

In the process:
1. Write a test
2. Make the test pass
3. Refactor

Sure, makes sense for development. As I spend 99% of my time reviewing and testing these days, I was just relaying my pain points :-)
 

in between each step you run the test more then 1 time, and depending on the refactoring you might need to run it several times. So imagine waiting 30 seconds per run to get results. To run a subset of tests is a pain because you need to be always changing the way you run the tests.....

I believe we could archive a better granularity and choosing what test to run if we used a runner like pytest or nose to do it. What was the reason behind handrolling a test runner script? I am asking this because in a previous job I decided to handroll a unittest loader script and that was something that I regretted every time I had to touch it, and eventually was in the process of changing it to pytest.

Pure newbie-ism. I have no objections to changing to something else, if it reduces our tech debt.
 

I looked into pytest to replace the current the current runtest, and the major problem I found was the testscenarios integration(See Note 1). It can be done but we would need to change all the test functions to receive the scenario variables through arguments on the function. Also didn't dug much into setting all the variables that we need there and all the environment.
The other issue that I do not like very much about pytest is the fact that you loose the unittest assertion that is not so bad because there are some neat libraries like: https://github.com/grappa-py/grappahttps://github.com/ActivisionGameScience/assertpyhttps://github.com/dgilland/verify. Personally I really like the syntax of Grapa, but the Veridfy one is pretty similar to Jasmine too.

What are your thoughts?

Huh, I also really like the grappa syntax. It's nice and readable.
 



Note 1: As an example of what our functions would have to look like you can see: https://github.com/OriMenashe/pytest-scenario/blob/master/tests/test_parametrize.py
As a example this class:

Without a diff, it's hard to be sure, but it looks like the only change was BaseTestGenerator to object on the first line?
 
class ServersWithServiceIDAddTestCase(BaseTestGenerator):
""" This class will add the servers under default server group. """

scenarios = [
# Fetch the default url for server object
(
'Default Server Node url', dict(
url='/browser/server/obj/'
)
)
]

def setUp(self):
pass

def runTest(self):
""" This function will add the server under default server group."""
url = "{0}{1}/".format(self.url, utils.SERVER_GROUP)
# Add service name in the config
self.server['service'] = "TestDB"
response = self.tester.post(
url,
data=json.dumps(self.server),
content_type='html/json'
)
self.assertEquals(response.status_code, 200)
response_data = json.loads(response.data.decode('utf-8'))
self.server_id = response_data['node']['_id']

def tearDown(self):
"""This function delete the server from SQLite """
utils.delete_server_with_api(self.tester, self.server_id)
Would have to look changed to:
class ServersWithServiceIDAddTestCase(object):
""" This class will add the servers under default server group. """

scenarios = [
# Fetch the default url for server object
(
'Default Server Node url', dict(
url='/browser/server/obj/'
)
)
]

def setUp(self):
pass

def runTest(self, url):
""" This function will add the server under default server group."""
url = "{0}{1}/".format(url, utils.SERVER_GROUP)
# Add service name in the config
self.server['service'] = "TestDB"
response = self.tester.post(
url,
data=json.dumps(self.server),
content_type='html/json'
)
self.assertEquals(response.status_code, 200)
response_data = json.loads(response.data.decode('utf-8'))
self.server_id = response_data['node']['_id']

def tearDown(self):
"""This function delete the server from SQLite """
utils.delete_server_with_api(self.tester, self.server_id)


Thanks
Joao

On Fri, Mar 9, 2018 at 8:31 AM Dave Page <dpage@pgadmin.org> wrote:
On Thu, Mar 8, 2018 at 2:22 PM, Joao De Almeida Pereira <jdealmeidapereira@pivotal.io> wrote:
Hello Khushboo,
Completely forgot about this python "feature".......
Attached is the fix.

Thanks, applied.
 

Just as a side question, does anyone else feel the pain of wanting to run a single test using a IDE or the command line and not being able to?

Not really - the Python and JS tests are so quick I don't really care (and with the Python ones, I can execute for a single module for even more speed).

What I would *really* like, is the ability to run individual feature tests. That would be very valuable and save me a ton of time.

 
We an HandRolled the loader, and that as some implications. Did anyone try to use a different launcher like pytest or nose instead of the current runner? 
I understand that testscenarios is one of the problems we have if we want to move away from this way of running tests.
Any suggestion?

Thanks
Joao

On Wed, Mar 7, 2018 at 11:41 PM Khushboo Vashi <khushboo.vashi@enterprisedb.com> wrote:
Hi Joao,

In the test_start_running_query.py, 2 static methods (is_begin_required_for_sql_query and is_rollback_statement_required)
of StartRunningQuery class were used directly without @patch. Due to this, in all the cases, the original value of them doesn't restore.

To fix this, I have sent the patch in another thread, to restore its original state, but I wonder if we can use these methods with @patch.

Thanks,
Khushboo


On Fri, Feb 9, 2018 at 5:24 PM, Dave Page <dpage@pgadmin.org> wrote:
Support EXPLAIN on Greenplum. Fixes #3097

 - Extract SQLEditor.execute and SQLEditor._poll into their own files and add test around them
 - Extract SQLEditor backend functions that start executing query to their own files and add tests around it
 - Move the Explain SQL from the front-end and now pass the Explain plan parameters as a JSON object in the start query call.
 - Extract the compile_template_name into a function that can be used by the different places that try to select the version of the template and the server type

Branch
------
master

Details
-------
https://git.postgresql.org/gitweb?p=pgadmin4.git;a=commitdiff;h=e16a95275336529a734bf0066889e39cc8ef0662
Author: Joao Pedro De Almeida Pereira <jdealmeidapereira@pivotal.io>

Modified Files
--------------
.../databases/schemas/tables/tests/test_utils.py   |    0
web/pgadmin/static/js/sqleditor/execute_query.js   |  287 ++++
.../js/sqleditor/is_new_transaction_required.js    |   14 +
.../static/js/sqleditor/query_tool_actions.js      |   33 +-
web/pgadmin/tools/sqleditor/__init__.py            |  396 +----
web/pgadmin/tools/sqleditor/static/js/sqleditor.js |  227 +--
.../sqleditor/sql/10_plus/explain_plan.sql         |   23 +
.../sqleditor/sql/9.2_plus/explain_plan.sql        |   20 +
.../sqleditor/sql/default/explain_plan.sql         |   17 +
.../sqleditor/sql/gpdb_5.0_plus/explain_plan.sql   |    5 +
web/pgadmin/tools/sqleditor/tests/__init__.py      |    8 +
.../sqleditor/tests/test_explain_plan_templates.py |  152 ++
.../test_extract_sql_from_network_parameters.py    |   59 +
.../tools/sqleditor/tests/test_start_query_tool.py |   38 +
web/pgadmin/tools/sqleditor/utils/__init__.py      |   14 +
.../sqleditor/utils/apply_explain_plan_wrapper.py  |   24 +
.../tools/sqleditor/utils/constant_definition.py   |   32 +
.../tools/sqleditor/utils/is_begin_required.py     |  169 ++
.../tools/sqleditor/utils/start_running_query.py   |  172 ++
.../tools/sqleditor/utils/tests/__init__.py        |    8 +
.../utils/tests/test_apply_explain_plan_wrapper.py |  121 ++
.../utils/tests/test_start_running_query.py        |  445 +++++
.../utils/update_session_grid_transaction.py       |   18 +
web/pgadmin/utils/compile_template_name.py         |   17 +
.../utils/tests/test_compile_template_name.py      |   34 +
web/pgadmin/utils/versioned_template_loader.py     |    2 +-
web/regression/javascript/fake_endpoints.js        |    6 +-
.../javascript/sqleditor/execute_query_spec.js     | 1702 ++++++++++++++++++++
.../sqleditor/is_new_transaction_required_spec.js  |   65 +
.../sqleditor/query_tool_actions_spec.js           |  141 +-
30 files changed, 3670 insertions(+), 579 deletions(-)





--
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake

EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company



--
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake

EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company

В списке pgadmin-hackers по дате отправления:

Предыдущее
От: Murtuza Zabuawala
Дата:
Сообщение: Re: [pgAdmin4][RM#3140] Add service parameter
Следующее
От: pgAdmin 4 Jenkins
Дата:
Сообщение: Build failed in Jenkins: pgadmin4-master-python27-feature #14