Adds DASH and PORT env vars

[OwenMatsuda]

This PR adds the ability to set a DASH_HOST environment variable to change the host value when running the server.

[alexcjohnson]

Alright, after way too much discussion with @josegonzalez and @rpkyle I'm on board with this change. Looks like you just have to adjust the usage re the test failure:

dash/dash.py:1913:13: W1508: os.getenv default type is builtins.int. Expected str or None. (invalid-envvar-default)

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int. Then we just need a CHANGELOG entry for the new options and I think this will be ready to 💃

[OwenMatsuda]

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int.

@alexcjohnson What kind of helpful behavior do you think? Should it just default to the 8050 port, or should it throw an error?

[alexcjohnson]

Throw an error like "Expecting an integer from 1 to 65535, found {}".format(repr(port)) (only including the range if you actually verify that, which would be a nice touch)

[OwenMatsuda]

I think this is ready for review. My only potential concern is that for port, when the user inputs it as a parameter within app.runserver(port=8050), port is an int. However, when grabbing it from an environment variable, port is a string that is later coerced into an int within the try/except block.

[alexcjohnson]

Just mark port here as either an int or a string containing an int, and we should be covered re: data types. Your coercion and try/except below looks great!

Can you please add a couple of tests of the failure cases, and maybe one of setting an integer string port and verifying that an app shows up there? These can probably all go in https://github.com/plotly/dash/blob/dev/tests/unit/test_configs.py

The failures can just be something like:

app=Dash()
app.layout=html.Div()
withpytest.raises(...) asexcinfo:
app.run_server(port="garbage")
assertexcinfo.value== ...

And the pass case can be something like:

app=Dash()
app.layout=html.Div("hi", id="out")
dash_duo.run_server(app, port="12345")
assertdash_duo.server_url=="https://127.0.0.1:12345"dash_duo.wait_for_text_to_equal("out", "hi")

[OwenMatsuda]

I'm trying to use pytest's monkeypatch to mock env vars, but I'm running into a bug: pytest-dev/pytest#6858

[alexcjohnson]

There's a fixture empty_environ at the top of the test_configs file - does that help?

[OwenMatsuda]

I saw that, but that would mean adding the env vars HOST AND PORT to dash/_configs.py. At first, I thought that would make more sense than the setup I have now, but then I noticed a problem:

env=load_dash_env_vars().get('DASH_{}'.format(name.upper()))
ifenvisNone:
returndefault

Unless we wanted to add another function specific to HOST and PORT, we would have to rename them to DASH_HOST, and DASH_PORT, respectively, which would be misaligned with plotly/dashR#167.

Along those same lines, I was wondering what the "parity" with Dash for Python was since the HOST and PORT didn't previously exist.

[alexcjohnson]

Beautiful. 💃

[OwenMatsuda]

This PR adds the ability to set a DASH_HOST environment variable to change the host value when running the server.

[alexcjohnson]

Alright, after way too much discussion with @josegonzalez and @rpkyle I'm on board with this change. Looks like you just have to adjust the usage re the test failure:

dash/dash.py:1913:13: W1508: os.getenv default type is builtins.int. Expected str or None. (invalid-envvar-default)

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int. Then we just need a CHANGELOG entry for the new options and I think this will be ready to 💃

[OwenMatsuda]

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int.

@alexcjohnson What kind of helpful behavior do you think? Should it just default to the 8050 port, or should it throw an error?

[alexcjohnson]

Throw an error like "Expecting an integer from 1 to 65535, found {}".format(repr(port)) (only including the range if you actually verify that, which would be a nice touch)

[OwenMatsuda]

I think this is ready for review. My only potential concern is that for port, when the user inputs it as a parameter within app.runserver(port=8050), port is an int. However, when grabbing it from an environment variable, port is a string that is later coerced into an int within the try/except block.

[alexcjohnson]

Just mark port here as either an int or a string containing an int, and we should be covered re: data types. Your coercion and try/except below looks great!

Can you please add a couple of tests of the failure cases, and maybe one of setting an integer string port and verifying that an app shows up there? These can probably all go in https://github.com/plotly/dash/blob/dev/tests/unit/test_configs.py

The failures can just be something like:

app=Dash()
app.layout=html.Div()
withpytest.raises(...) asexcinfo:
app.run_server(port="garbage")
assertexcinfo.value== ...

And the pass case can be something like:

app=Dash()
app.layout=html.Div("hi", id="out")
dash_duo.run_server(app, port="12345")
assertdash_duo.server_url=="https://127.0.0.1:12345"dash_duo.wait_for_text_to_equal("out", "hi")

[OwenMatsuda]

I'm trying to use pytest's monkeypatch to mock env vars, but I'm running into a bug: pytest-dev/pytest#6858

[alexcjohnson]

There's a fixture empty_environ at the top of the test_configs file - does that help?

[OwenMatsuda]

I saw that, but that would mean adding the env vars HOST AND PORT to dash/_configs.py. At first, I thought that would make more sense than the setup I have now, but then I noticed a problem:

env=load_dash_env_vars().get('DASH_{}'.format(name.upper()))
ifenvisNone:
returndefault

Unless we wanted to add another function specific to HOST and PORT, we would have to rename them to DASH_HOST, and DASH_PORT, respectively, which would be misaligned with plotly/dashR#167.

Along those same lines, I was wondering what the "parity" with Dash for Python was since the HOST and PORT didn't previously exist.

[alexcjohnson]

Beautiful. 💃

[OwenMatsuda]

This PR adds the ability to set a DASH_HOST environment variable to change the host value when running the server.

[alexcjohnson]

Alright, after way too much discussion with @josegonzalez and @rpkyle I'm on board with this change. Looks like you just have to adjust the usage re the test failure:

dash/dash.py:1913:13: W1508: os.getenv default type is builtins.int. Expected str or None. (invalid-envvar-default)

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int. Then we just need a CHANGELOG entry for the new options and I think this will be ready to 💃

[OwenMatsuda]

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int.

@alexcjohnson What kind of helpful behavior do you think? Should it just default to the 8050 port, or should it throw an error?

[alexcjohnson]

Throw an error like "Expecting an integer from 1 to 65535, found {}".format(repr(port)) (only including the range if you actually verify that, which would be a nice touch)

[OwenMatsuda]

I think this is ready for review. My only potential concern is that for port, when the user inputs it as a parameter within app.runserver(port=8050), port is an int. However, when grabbing it from an environment variable, port is a string that is later coerced into an int within the try/except block.

[alexcjohnson]

Just mark port here as either an int or a string containing an int, and we should be covered re: data types. Your coercion and try/except below looks great!

Can you please add a couple of tests of the failure cases, and maybe one of setting an integer string port and verifying that an app shows up there? These can probably all go in https://github.com/plotly/dash/blob/dev/tests/unit/test_configs.py

The failures can just be something like:

app=Dash()
app.layout=html.Div()
withpytest.raises(...) asexcinfo:
app.run_server(port="garbage")
assertexcinfo.value== ...

And the pass case can be something like:

app=Dash()
app.layout=html.Div("hi", id="out")
dash_duo.run_server(app, port="12345")
assertdash_duo.server_url=="https://127.0.0.1:12345"dash_duo.wait_for_text_to_equal("out", "hi")

[OwenMatsuda]

I'm trying to use pytest's monkeypatch to mock env vars, but I'm running into a bug: pytest-dev/pytest#6858

[alexcjohnson]

There's a fixture empty_environ at the top of the test_configs file - does that help?

[OwenMatsuda]

I saw that, but that would mean adding the env vars HOST AND PORT to dash/_configs.py. At first, I thought that would make more sense than the setup I have now, but then I noticed a problem:

env=load_dash_env_vars().get('DASH_{}'.format(name.upper()))
ifenvisNone:
returndefault

Unless we wanted to add another function specific to HOST and PORT, we would have to rename them to DASH_HOST, and DASH_PORT, respectively, which would be misaligned with plotly/dashR#167.

Along those same lines, I was wondering what the "parity" with Dash for Python was since the HOST and PORT didn't previously exist.

[alexcjohnson]

Beautiful. 💃

[OwenMatsuda]

This PR adds the ability to set a DASH_HOST environment variable to change the host value when running the server.

[alexcjohnson]

Alright, after way too much discussion with @josegonzalez and @rpkyle I'm on board with this change. Looks like you just have to adjust the usage re the test failure:

dash/dash.py:1913:13: W1508: os.getenv default type is builtins.int. Expected str or None. (invalid-envvar-default)

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int. Then we just need a CHANGELOG entry for the new options and I think this will be ready to 💃

[OwenMatsuda]

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int.

@alexcjohnson What kind of helpful behavior do you think? Should it just default to the 8050 port, or should it throw an error?

[alexcjohnson]

Throw an error like "Expecting an integer from 1 to 65535, found {}".format(repr(port)) (only including the range if you actually verify that, which would be a nice touch)

[OwenMatsuda]

I think this is ready for review. My only potential concern is that for port, when the user inputs it as a parameter within app.runserver(port=8050), port is an int. However, when grabbing it from an environment variable, port is a string that is later coerced into an int within the try/except block.

[alexcjohnson]

Just mark port here as either an int or a string containing an int, and we should be covered re: data types. Your coercion and try/except below looks great!

Can you please add a couple of tests of the failure cases, and maybe one of setting an integer string port and verifying that an app shows up there? These can probably all go in https://github.com/plotly/dash/blob/dev/tests/unit/test_configs.py

The failures can just be something like:

app=Dash()
app.layout=html.Div()
withpytest.raises(...) asexcinfo:
app.run_server(port="garbage")
assertexcinfo.value== ...

And the pass case can be something like:

app=Dash()
app.layout=html.Div("hi", id="out")
dash_duo.run_server(app, port="12345")
assertdash_duo.server_url=="https://127.0.0.1:12345"dash_duo.wait_for_text_to_equal("out", "hi")

[OwenMatsuda]

I'm trying to use pytest's monkeypatch to mock env vars, but I'm running into a bug: pytest-dev/pytest#6858

[alexcjohnson]

There's a fixture empty_environ at the top of the test_configs file - does that help?

[OwenMatsuda]

I saw that, but that would mean adding the env vars HOST AND PORT to dash/_configs.py. At first, I thought that would make more sense than the setup I have now, but then I noticed a problem:

env=load_dash_env_vars().get('DASH_{}'.format(name.upper()))
ifenvisNone:
returndefault

Unless we wanted to add another function specific to HOST and PORT, we would have to rename them to DASH_HOST, and DASH_PORT, respectively, which would be misaligned with plotly/dashR#167.

Along those same lines, I was wondering what the "parity" with Dash for Python was since the HOST and PORT didn't previously exist.

[alexcjohnson]

Beautiful. 💃

[OwenMatsuda]

This PR adds the ability to set a DASH_HOST environment variable to change the host value when running the server.

[alexcjohnson]

Alright, after way too much discussion with @josegonzalez and @rpkyle I'm on board with this change. Looks like you just have to adjust the usage re the test failure:

dash/dash.py:1913:13: W1508: os.getenv default type is builtins.int. Expected str or None. (invalid-envvar-default)

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int. Then we just need a CHANGELOG entry for the new options and I think this will be ready to 💃

[OwenMatsuda]

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int.

@alexcjohnson What kind of helpful behavior do you think? Should it just default to the 8050 port, or should it throw an error?

[alexcjohnson]

Throw an error like "Expecting an integer from 1 to 65535, found {}".format(repr(port)) (only including the range if you actually verify that, which would be a nice touch)

[OwenMatsuda]

I think this is ready for review. My only potential concern is that for port, when the user inputs it as a parameter within app.runserver(port=8050), port is an int. However, when grabbing it from an environment variable, port is a string that is later coerced into an int within the try/except block.

[alexcjohnson]

Just mark port here as either an int or a string containing an int, and we should be covered re: data types. Your coercion and try/except below looks great!

Can you please add a couple of tests of the failure cases, and maybe one of setting an integer string port and verifying that an app shows up there? These can probably all go in https://github.com/plotly/dash/blob/dev/tests/unit/test_configs.py

The failures can just be something like:

app=Dash()
app.layout=html.Div()
withpytest.raises(...) asexcinfo:
app.run_server(port="garbage")
assertexcinfo.value== ...

And the pass case can be something like:

app=Dash()
app.layout=html.Div("hi", id="out")
dash_duo.run_server(app, port="12345")
assertdash_duo.server_url=="https://127.0.0.1:12345"dash_duo.wait_for_text_to_equal("out", "hi")

[OwenMatsuda]

I'm trying to use pytest's monkeypatch to mock env vars, but I'm running into a bug: pytest-dev/pytest#6858

[alexcjohnson]

There's a fixture empty_environ at the top of the test_configs file - does that help?

[OwenMatsuda]

I saw that, but that would mean adding the env vars HOST AND PORT to dash/_configs.py. At first, I thought that would make more sense than the setup I have now, but then I noticed a problem:

env=load_dash_env_vars().get('DASH_{}'.format(name.upper()))
ifenvisNone:
returndefault

Unless we wanted to add another function specific to HOST and PORT, we would have to rename them to DASH_HOST, and DASH_PORT, respectively, which would be misaligned with plotly/dashR#167.

Along those same lines, I was wondering what the "parity" with Dash for Python was since the HOST and PORT didn't previously exist.

[alexcjohnson]

Beautiful. 💃

[OwenMatsuda]

This PR adds the ability to set a DASH_HOST environment variable to change the host value when running the server.

[alexcjohnson]

Alright, after way too much discussion with @josegonzalez and @rpkyle I'm on board with this change. Looks like you just have to adjust the usage re the test failure:

dash/dash.py:1913:13: W1508: os.getenv default type is builtins.int. Expected str or None. (invalid-envvar-default)

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int. Then we just need a CHANGELOG entry for the new options and I think this will be ready to 💃

[OwenMatsuda]

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int.

@alexcjohnson What kind of helpful behavior do you think? Should it just default to the 8050 port, or should it throw an error?

[alexcjohnson]

Throw an error like "Expecting an integer from 1 to 65535, found {}".format(repr(port)) (only including the range if you actually verify that, which would be a nice touch)

[OwenMatsuda]

I think this is ready for review. My only potential concern is that for port, when the user inputs it as a parameter within app.runserver(port=8050), port is an int. However, when grabbing it from an environment variable, port is a string that is later coerced into an int within the try/except block.

[alexcjohnson]

Just mark port here as either an int or a string containing an int, and we should be covered re: data types. Your coercion and try/except below looks great!

Can you please add a couple of tests of the failure cases, and maybe one of setting an integer string port and verifying that an app shows up there? These can probably all go in https://github.com/plotly/dash/blob/dev/tests/unit/test_configs.py

The failures can just be something like:

app=Dash()
app.layout=html.Div()
withpytest.raises(...) asexcinfo:
app.run_server(port="garbage")
assertexcinfo.value== ...

And the pass case can be something like:

app=Dash()
app.layout=html.Div("hi", id="out")
dash_duo.run_server(app, port="12345")
assertdash_duo.server_url=="https://127.0.0.1:12345"dash_duo.wait_for_text_to_equal("out", "hi")

[OwenMatsuda]

I'm trying to use pytest's monkeypatch to mock env vars, but I'm running into a bug: pytest-dev/pytest#6858

[alexcjohnson]

There's a fixture empty_environ at the top of the test_configs file - does that help?

[OwenMatsuda]

I saw that, but that would mean adding the env vars HOST AND PORT to dash/_configs.py. At first, I thought that would make more sense than the setup I have now, but then I noticed a problem:

env=load_dash_env_vars().get('DASH_{}'.format(name.upper()))
ifenvisNone:
returndefault

Unless we wanted to add another function specific to HOST and PORT, we would have to rename them to DASH_HOST, and DASH_PORT, respectively, which would be misaligned with plotly/dashR#167.

Along those same lines, I was wondering what the "parity" with Dash for Python was since the HOST and PORT didn't previously exist.

[alexcjohnson]

Beautiful. 💃

[OwenMatsuda]

This PR adds the ability to set a DASH_HOST environment variable to change the host value when running the server.

[alexcjohnson]

Alright, after way too much discussion with @josegonzalez and @rpkyle I'm on board with this change. Looks like you just have to adjust the usage re the test failure:

dash/dash.py:1913:13: W1508: os.getenv default type is builtins.int. Expected str or None. (invalid-envvar-default)

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int. Then we just need a CHANGELOG entry for the new options and I think this will be ready to 💃

[OwenMatsuda]

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int.

@alexcjohnson What kind of helpful behavior do you think? Should it just default to the 8050 port, or should it throw an error?

[alexcjohnson]

Throw an error like "Expecting an integer from 1 to 65535, found {}".format(repr(port)) (only including the range if you actually verify that, which would be a nice touch)

[OwenMatsuda]

I think this is ready for review. My only potential concern is that for port, when the user inputs it as a parameter within app.runserver(port=8050), port is an int. However, when grabbing it from an environment variable, port is a string that is later coerced into an int within the try/except block.

[alexcjohnson]

Just mark port here as either an int or a string containing an int, and we should be covered re: data types. Your coercion and try/except below looks great!

Can you please add a couple of tests of the failure cases, and maybe one of setting an integer string port and verifying that an app shows up there? These can probably all go in https://github.com/plotly/dash/blob/dev/tests/unit/test_configs.py

The failures can just be something like:

app=Dash()
app.layout=html.Div()
withpytest.raises(...) asexcinfo:
app.run_server(port="garbage")
assertexcinfo.value== ...

And the pass case can be something like:

app=Dash()
app.layout=html.Div("hi", id="out")
dash_duo.run_server(app, port="12345")
assertdash_duo.server_url=="https://127.0.0.1:12345"dash_duo.wait_for_text_to_equal("out", "hi")

[OwenMatsuda]

I'm trying to use pytest's monkeypatch to mock env vars, but I'm running into a bug: pytest-dev/pytest#6858

[alexcjohnson]

There's a fixture empty_environ at the top of the test_configs file - does that help?

[OwenMatsuda]

I saw that, but that would mean adding the env vars HOST AND PORT to dash/_configs.py. At first, I thought that would make more sense than the setup I have now, but then I noticed a problem:

env=load_dash_env_vars().get('DASH_{}'.format(name.upper()))
ifenvisNone:
returndefault

Unless we wanted to add another function specific to HOST and PORT, we would have to rename them to DASH_HOST, and DASH_PORT, respectively, which would be misaligned with plotly/dashR#167.

Along those same lines, I was wondering what the "parity" with Dash for Python was since the HOST and PORT didn't previously exist.

[alexcjohnson]

Beautiful. 💃

[OwenMatsuda]

This PR adds the ability to set a DASH_HOST environment variable to change the host value when running the server.

[alexcjohnson]

Alright, after way too much discussion with @josegonzalez and @rpkyle I'm on board with this change. Looks like you just have to adjust the usage re the test failure:

dash/dash.py:1913:13: W1508: os.getenv default type is builtins.int. Expected str or None. (invalid-envvar-default)

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int. Then we just need a CHANGELOG entry for the new options and I think this will be ready to 💃

[OwenMatsuda]

And let's give it some helpful behavior if you fail to provide an int or a string that can be coerced to an int.

@alexcjohnson What kind of helpful behavior do you think? Should it just default to the 8050 port, or should it throw an error?

[alexcjohnson]

Throw an error like "Expecting an integer from 1 to 65535, found {}".format(repr(port)) (only including the range if you actually verify that, which would be a nice touch)

[OwenMatsuda]

I think this is ready for review. My only potential concern is that for port, when the user inputs it as a parameter within app.runserver(port=8050), port is an int. However, when grabbing it from an environment variable, port is a string that is later coerced into an int within the try/except block.

[alexcjohnson]

Just mark port here as either an int or a string containing an int, and we should be covered re: data types. Your coercion and try/except below looks great!

Can you please add a couple of tests of the failure cases, and maybe one of setting an integer string port and verifying that an app shows up there? These can probably all go in https://github.com/plotly/dash/blob/dev/tests/unit/test_configs.py

The failures can just be something like:

app=Dash()
app.layout=html.Div()
withpytest.raises(...) asexcinfo:
app.run_server(port="garbage")
assertexcinfo.value== ...

And the pass case can be something like:

app=Dash()
app.layout=html.Div("hi", id="out")
dash_duo.run_server(app, port="12345")
assertdash_duo.server_url=="https://127.0.0.1:12345"dash_duo.wait_for_text_to_equal("out", "hi")

[OwenMatsuda]

I'm trying to use pytest's monkeypatch to mock env vars, but I'm running into a bug: pytest-dev/pytest#6858

[alexcjohnson]

There's a fixture empty_environ at the top of the test_configs file - does that help?

[OwenMatsuda]

I saw that, but that would mean adding the env vars HOST AND PORT to dash/_configs.py. At first, I thought that would make more sense than the setup I have now, but then I noticed a problem:

env=load_dash_env_vars().get('DASH_{}'.format(name.upper()))
ifenvisNone:
returndefault

Unless we wanted to add another function specific to HOST and PORT, we would have to rename them to DASH_HOST, and DASH_PORT, respectively, which would be misaligned with plotly/dashR#167.

Along those same lines, I was wondering what the "parity" with Dash for Python was since the HOST and PORT didn't previously exist.

[alexcjohnson]

Beautiful. 💃