Automatically move position

[snarlysodboxer]

1, I like annotations after my models.
2, I frequently forget to add the '-p after' option when using this gem.
3, Simply running the command again adding the '-p after' option does not change the position since it detects that the model has already been annotated.
4, During writing this functionality I discovered you could simply pass the --force option, however there is nothing in the documentation about this, and it seems to me this functionality should be default, so I wrote the change anyway.

I understand if other people feel differently about moving the position of the annotations by default.

Much like the other code I see in annotate_models, to detect the need to move the position of the annotations, this code relies on the following two regular expressions to find the beginning and end of the string (model file):

/\A# == Schema Info/
/\n#\n\n\z/

[alexch]

Auto-detecting position is a feature. Doesn't this remove that feature? Maybe instead we should just document better that -p defers to the current position unless --force is set.

On Apr 6, 2013, at 1:04 PM, David notifications@github.com wrote:

1, I like annotations after my models.
2, I frequently forget to add the '-p after' option when using this gem.
3, Simply running the command again adding the '-p after' option does not change the position since it detects that the model has already been annotated.
4, During writing this functionality I discovered you could simply pass the --force option, however there is nothing in the documentation about this, and it seems to me this functionality should be default, so I wrote the change anyway.

I understand if other people feel differently about moving the position of the annotations by default.

Much like the other code I see in annotate_models, to detect the need to move the position of the annotations, this code relies on the following two regular expressions to find the beginning and end of the string (model file):

/\A# == Schema Info/
/\n#\n\n\z/

You can merge this Pull Request by running

git pull https://github.com/snarlysodboxer/annotate_models master
Or view, comment on, or merge it at:

#119

Commit Summary

Automatically move position
File Changes

M lib/annotate/annotate_models.rb (9)
M spec/annotate/annotate_models_spec.rb (19)
Patch Links:

https://github.com/ctran/annotate_models/pull/119.patch
https://github.com/ctran/annotate_models/pull/119.diff

[namick]

If I understand this correctly, this does not remove auto-detection of the position, rather, it continues to auto-detect unless overridden by -p.

Currently -p is ignored if the schema has not changed. This seems to fix that.

+1 for me.

[snarlysodboxer]

Correct me if I'm missing something here guys, but if you read the code you will find nothing that does any "detecting the position" in any way. (The two regex's I listed in this pull request would be this code base's first and only 'position-detection' code if they get accepted.) Let me explain for you:

As ctran/annotate_models stands right now:
Background: a model file is loaded into a variable that is a string.

1. Annotate_models runs regular expressions on that string to detect any annotations it previously added, regardless of the position of those annotations. Those old annotations are put into a variable called old_columns. (Understand this to mean old_content, or old_annotations.)
2. Annotate_models runs it's code that determines what the new_columns should be.
3. It then compares old_columns to new_columns and:

• If they're the same, it does nothing,
• If they're different, it uses ruby's sub! command to replace old_columns with an empty string (regardless of old_column's position in the file), and then it continues on to again use sub! to replace an empty string at whatever-position-was-chosen-this-run with the content of new_columns.

Effectively, this means:
Given I have ten models and all of them are previously annotated with the -p after option.

1. I run a migration that adds one column to one of my tables.
2. I run annotate (forgetting to add the -p after option)
3. I end up with 9 models annotated after, and 1 model annotated before.
   OK, no big deal:
4. I realize my mistake, so I re-run it correctly annotate -p after
5. No changes are made because the files have all already been annotated with the current schema. Doh!
6. I then run annotate -p after --force and all ten of the models will be re-annotated, which is unnecessary and changes the mtime on all of the files, and that could lead to other annoying side effects.

Regardless of forgetting to add the -p after option the first time, doesn't it make sense that if I run the annotate command, then later run annotate -p after it should move the annotations to the after position? (And vise versa.) - This code provides that.

[ctran]

This now works as you expected in 2.6.2 by retaining the current position unless you pass "--force" option.