default_values

[benjello]

This a a non working tentative which tries to reimplement @AlexisEidelman default_value declaration
@gdementen if you have a bit of time to have look at it I would be very glad.
And if you think, that it I am in the wrong direction, please be blunt !
Thanks

[landscape-bot]

Repository health decreased by 0.75% when pulling 045a689 on benjello:defaut_values into 7709f8f on liam2:master.

• 13 new problems were found (including 5 errors and 4 code smells).
• 1 problem was fixed (including 0 errors and 0 code smells).

[gdementen]

empty should accept an optional arg, and this line should use it instead

[gdementen]

There are strange things in this proposition, but it's hard to tell what exactly without thinking hard about it and I am too tired to do it today and will probably not have time to do it for a few weeks.

[benjello]

Thanks @gdementen, it is messy because it is mainly a merge with a few fix. I will try on my own for a moment.

[landscape-bot]

Repository health decreased by 0.25% when pulling 7a31efc on benjello:defaut_values into 7709f8f on liam2:master.

• 8 new problems were found (including 2 errors and 1 code smell).
• 2 problems were fixed (including 0 errors and 1 code smell).

[benjello]

@gdementen : should I modify the get_missing_* by adding a default_values argument or should I add it only to ColumnArray.empty and the calls to this method in ColumnArray.from_table and ColumnArray.from_records ?

[gdementen]

@benjello I don't know. You should see what works best. I would find it elegant if we could suppress the concept of missing value, and only have default values which themselves default to the current values we use for missing values depending on the type of field.

[benjello]

It is somehow what we have done in openfisca and it is actually useful. I
will try but I may come with more questions ;-) so stop when you get bored

Le jeu. 29 oct. 2015 09:03, Gaëtan de Menten notifications@github.com a
écrit :

@benjello https://github.com/benjello I don't know. You should see what
works best. I would find it elegant if we could suppress the concept of
missing value, and only have default values which themselves default to the
current values we use for missing values depending on the type of field.

—
Reply to this email directly or view it on GitHub
#190 (comment).

[gdementen]

I won't get bored, but I am low on time... On the other hand I want to avoid slowing you down or discouraging you from contributing to LIAM2.

[benjello]

We are all low on time ;-)

Le jeu. 29 oct. 2015 09:17, Gaëtan de Menten notifications@github.com a
écrit :

I won't get bored, but I am low on time... On the other hand I want to
avoid slowing you down or discouraging you from contributing to LIAM2.

—
Reply to this email directly or view it on GitHub
#190 (comment).

[gdementen]

FWIW, np.full is numpy1.8+, so we'd need to bump the requirements in INSTALL.

[landscape-bot]

Repository health decreased by 0.64% when pulling 4931d95 on benjello:defaut_values into 7344808 on liam2:master.

• 14 new problems were found (including 6 errors and 2 code smells).
• 4 problems were fixed (including 0 errors and 2 code smells).

[landscape-bot]

Repository health decreased by 0.64% when pulling 79e1e29 on benjello:defaut_values into 7344808 on liam2:master.

• 14 new problems were found (including 6 errors and 2 code smells).
• 4 problems were fixed (including 0 errors and 2 code smells).

[benjello]

I introduced default_values everywhere the was a missing_* but still it the test do not pass. I confess I do not figure the entire process very well so i could use a hint about how to debug this kind of problem ;-)

[landscape-bot]

Repository health decreased by 0.64% when pulling 40bac28 on benjello:defaut_values into 7344808 on liam2:master.

• 14 new problems were found (including 6 errors and 2 code smells).
• 4 problems were fixed (including 0 errors and 2 code smells).

[gdementen]

The line "defaulted_integer_variable: [99 99 99 ..., -1 -1 -1] done (0 ms elapsed)." makes me suspect the new borns do not use the default value, which means the thing is probably working as it should for the initial population but not for newborns. Did you implement it?

[benjello]

Good catch @gdementen. Tests do pass now. I can add more tests if needed.
You should carefully review this PR, i was not always very confident about what i was modifying.
I still lack a clear vision of the global architecture.

[landscape-bot]

Repository health decreased by 0.68% when pulling 3b05758 on benjello:defaut_values into 7344808 on liam2:master.

• 15 new problems were found (including 6 errors and 3 code smells).
• 4 problems were fixed (including 0 errors and 2 code smells).

[gdementen]

I will not have time for a in-depth review & merge for a couple of weeks, but in the mean time, could you;

• provide documentation
• change the default value for the default arguments to None instead of {} (and update the code accordingly). Using a mutable default value for an argument is asking for trouble.
• ideally add a mention in the changelog at doc/usersguide/source/changes/version_0_11.rst.inc
  Thanks!

[benjello]

I wil try to find some time for that.

[landscape-bot]

Repository health decreased by 0.07% when pulling dab7c8b on benjello:defaut_values into 58bea98 on liam2:master.

• 9 new problems were found (including 0 errors and 3 code smells).
• 4 problems were fixed (including 0 errors and 2 code smells).

[gdementen]

this is not True (or at least was not True -- and should not be). -1 for integer and 'nan' for float

[gdementen]

You do no speak about new(). In fact I think a separate paragraph and saying it applies to both new() and fields not present in the input bla bla... would be better, what do you think?

[benjello]

You are right. Will do that.

On Thu, Nov 12, 2015 at 10:43 PM Gaëtan de Menten notifications@github.com
wrote:

In doc/usersguide/source/model.rst
#190 (comment):

@@ -219,6 +219,10 @@ However, in practice, there are often some fields which are not present in the
input file. They will need to be calculated later by the model, and you need to
tell LIAM2 that the field is missing, by using initialdata: False in the
definition for that field (see the agegroup variable in the example below).
+The fields that are not present in the initial file can also be initialized to
+a specific value by using the default: some_default_value. If not present they
+are initialized to the default of the field type which are False for boolean and
+0 for integer and float (see the alive variable in the example below).

You do no speak about new(). In fact I think a separate paragraph and
saying it applies to both new() and fields not present in the input bla
bla... would be better, what do you think?

—
Reply to this email directly or view it on GitHub
https://github.com/liam2/liam2/pull/190/files#r44719011.

[landscape-bot]

Repository health decreased by 0.07% when pulling 8535c9e on benjello:defaut_values into 58bea98 on liam2:master.

• 9 new problems were found (including 0 errors and 3 code smells).
• 4 problems were fixed (including 0 errors and 2 code smells).

[landscape-bot]

Repository health decreased by 0.07% when pulling 3e8083c on benjello:defaut_values into bfcae0a on liam2:master.

• 9 new problems were found (including 0 errors and 3 code smells).
• 4 problems were fixed (including 0 errors and 2 code smells).

[landscape-bot]

Repository health decreased by 0.07% when pulling 761324f on benjello:defaut_values into bfcae0a on liam2:master.

• 9 new problems were found (including 0 errors and 3 code smells).
• 4 problems were fixed (including 0 errors and 2 code smells).

[landscape-bot]

Repository health decreased by 0.17% when pulling e5b4198 on benjello:defaut_values into 9162627 on liam2:master.

• 9 new problems were found (including 1 error and 2 code smells).
• 3 problems were fixed (including 0 errors and 1 code smell).

[gdementen]

Merged! Thanks a lot!

[benjello]

I hope you checked it carefully since I was not always confortable with what I was modifying.

[gdementen]

Don't worry, I did not merge it as-is. I did check & modify it. There is still a bug left, but I wanted to add a test case for it before fixing it.