Dynamic Records Meritfront

Dyanmic Records Meritfront helps extend active record functionality to make it more dynamic. These methods have the goal of allowing one to

  1. communicate with the frontend quicker and more effectively through Global HashIds
  2. communicate with the backend more effectively with sql queries. This becomes especially relevant when you hit the limits of Active Record Relations and the usual way of querying in rails. For instance, if you have dynamic sql queries that are hard to convert properly into ruby.
  3. add other helper methods to work with your database, such as checking if relations exist, or if a migration has been run.

Note that you will have issues using the gem outside of postgres. These issue can be fixed, and I am considering a major version update to fix them. Let me know and I will probably do it.

Basic Examples

# returns a json-like hash list of user data
users = ApplicationRecord.dynamic_sql(
    'select * from users limit :our_limit',
    our_limit: 5
)

#returns a list of users (each an instance of User)
users = User.dynamic_sql(
    'select * from users where id = ANY (:ids)',
    ids: [1,2,3]
)

uhgid = users.first.hgid        #returns a hashed global id like: 'gid://appname/User/K9YI4K'
user = User.locate_hgid(uhgid)      #returns that user

#... and much more!

Installation

Add this line to your application's Gemfile:

gem 'dynamic-records-meritfront'

And then execute:

$ bundle install

Or install it yourself as:

$ gem install dynamic-records-meritfront

Usage

Apply to your ApplicationRecord class as such (or whatever subclass of ActiveRecord::Base you have)

class ApplicationRecord < ActiveRecord::Base
    self.abstract_class = true
    include DynamicRecordsMeritfront
end

SQL methods

Methods written for easier sql usage.

self.dynamic_sql( optional name, sql, opts = { })

A better and safer way to write sql. Can return either a Hash, ActiveRecord::Response object, or an instantiated model.

User.dynamic_sql('select * from users') #returns all users
ApplicationRecord.dynamic_sql('select * from users') #returns all user column information in an array

with options:

  • options not stated below: considered sql arguments, and will replace their ":option_name" with a sql argument. Always use sql arguments to avoid sql injection. Lists are converted into a format such as {1,2,3,4}. Lists of lists are converted into (1,2,3), (4,5,6), (7,8,9) etc. So as to allow easy inserts/upserts.
  • prepare: Defaults to true. Gets passed to ActiveRecord::Base.connection.exec_query as a parameter. Should change whether the command will be prepared, which means that on subsequent calls the command will be faster. Downsides are when, for example, the sql query has hard-coded arguments, the query always changes, causing technical issues as the number of prepared statements stack up.
  • multi_query: allows more than one query (you can seperate an insert and an update with ';' I dont know how else to say it.) this disables other options including sql_arguments. Not sure how it effects prepared statements. Not super useful.
  • async: Defaults to false. Gets passed to ActiveRecord::Base.connection.exec_query as a parameter. See that methods documentation for more. I was looking through the source code, and I think it only effects how it logs to the logfile?
example usage Delete Friend Requests between two users after they have become friends. ```ruby ApplicationRecord.dynamic_sql("remove_friend_requests_before_creating_friend", %Q{ DELETE FROM friend_requests WHERE (requestie_id = :uid and requester_id = :other_user_id) OR (requester_id = :uid and requestie_id = :other_user_id) }, uid: Current.user.id, other_user_id: other_user.id) ```
example usage with interpreted sql string Get all users who have made a friend request to a particular user with an optional limit. This is an example of why this method is good for dynamic prepared statements. ```ruby return User.dynamic_sql('get_friend_requests', %Q{ SELECT * FROM ( SELECT other_user.id, ..., friend_requests.created_at FROM users INNER JOIN friend_requests ON users.id = friend_requests.requestie_id INNER JOIN users other_user ON friend_requests.requester_id = other_user.id WHERE users.id = :uid ) AS all_friend_requests ORDER BY all_friend_requests.created_at DESC #:limit" if limit > 0 }, uid: u, limit: limit) ```
access non-standard column for table with a ActiveRecord model ```ruby #get a normal test vote test = Vote.dynamic_sql(%Q{ SELECT id FROM votes LIMIT 1 }).first v.inspect # "#" #get a cool test vote. Note that is_this_vote_cool is not on the vote table. test = Vote.dynamic_sql(%Q{ SELECT id, 'yes' AS is_this_vote_cool FROM votes LIMIT 1 }).first test.inspect # # #getting attributes added dynamically to the models, and also showing up on inspects, was... more difficult than i anticipated. ```
example usage with selecting records that match list of ids Get users who match a list of ids. Uses a postgresql Array, see the potential issues section ```ruby id_list = [1,2,3] return User.dynamic_sql('get_usrs', %Q{ SELECT * FROM users WHERE id = ANY (:id_list) }, id_list: id_list) ```
example usage a custom upsert Do an upsert ```ruby time = DateTime.now rows = uzrs.map{|u| [ u.id, #user_id self.id, #conversation_id from, #invited_by :time, #created_at (We use symbols to denote other sql arguments) :time, #updated_at ]} ApplicationRecord.dynamic_sql("upsert_conversation_invites_2", %Q{ INSERT INTO conversation_participants (user_id, conversation_id, invited_by, created_at, updated_at) VALUES :rows ON CONFLICT (conversation_id,user_id) DO UPDATE SET updated_at = :time }, rows: rows, time: time) ``` This will output sql similar to below. Note this can be done for multiple conversation_participants. Also note that we sent only one time variable during our request instead of duplicating it. ```sql INSERT INTO conversation_participants (user_id, conversation_id, invited_by, created_at, updated_at) VALUES ($1,$2,$3,$4,$4) ON CONFLICT (conversation_id,user_id) DO UPDATE SET updated_at = $4 -- [["rows_1", 15], ["rows_2", 67], ["rows_3", 6], [:time, "2022-10-13 20:49:27.441372"]] ```

self.dynamic_preload(records, associations)

Preloads from a list of records, and not from a ActiveRecord_Relation. This will be useful when using the above dynamic_sql method (as it returns a list of records, and not a record relation). This is basically the same as a normal relation preload but it works on a list.

ApplicationRecord.dynamic_preload(comments, [:votes])
example usage Preload :votes on some comments. :votes is an active record has_many relation. ```ruby comments = Comment.dynamic_sql('get_comments', %Q{ SELECT * FROM comments LIMIT 4 }) comments.class.to_s # 'Array' note: not a relation. ApplicationRecord.dynamic_preload(comments, [:votes]) puts comments[0].votes #this line should be preloaded and hence not call the database #note that this above is basically the same as doing the below assuming there is a comments relation on the user model. user.comments.preload(:votes) ```

has_run_migration?(nm)

put in a string name of the migration's class and it will say if it has allready run the migration. good during enum migrations as the code to migrate wont run if enumerate is there as it is not yet enumerated (causing an error when it loads the class that will have the enumeration in it). This can lead it to being impossible to commit clean code.

ApplicationRecord.has_run_migration?('UserImageRelationsTwo')
example usage only load relationa if it exists in the database ```ruby if ApplicationRecord.has_run_migration?('UserImageRelationsTwo') class UserImageRelation < ApplicationRecord belongs_to :imageable, polymorphic: true belongs_to :image end else class UserImageRelation; end end ```

has_association?(*args)

accepts a list of association names, checks if the model has those associations

obj.has_association?(:votes)
example usage Check if object is a votable class ```ruby obj = Comment.first obj.has_association?(:votes) #true obj = User.first obj.has_association?(:votes) #false ```

self.instaload_sql( optional name, insta_array, opts = { })

instaloads a bunch of diffrent models at the same time by casting them to json before returning them. Kinda cool. Maybe a bit overcomplicated. Seems to be more efficient to preloading when i tested it.

  • name is passed to dynamic_sql and is the name of the sql request
  • opts are passed to dynamic_sql
  • requires a list of instaload method output which provides information for how to treat each sql block.
out = ApplicationRecord.instaload_sql([
    ApplicationRecord.instaload("SELECT id FROM users", relied_on: true, dont_return: true, table_name: "users_2"),
    ApplicationRecord.instaload("SELECT id FROM users_2 WHERE id % 2 != 0 LIMIT :limit", table_name: 'a'),
    User.instaload("SELECT id FROM users_2 WHERE id % 2 != 1 LIMIT :limit", table_name: 'b')
], limit: 2)
example usage #get list of users, those users friends, and who those users follow, all in one request. ```ruby # the ruby entered output = ApplicationRecord.instaload_sql([ User.instaload('SELECT id FROM users WHERE users.id = ANY (:user_ids) AND users.created_at > :time', table_name: 'limited_users', relied_on: true), User.instaload(%Q{ SELECT friends.smaller_user_id AS id, friends.bigger_user_id AS friended_to FROM friends INNER JOIN limited_users ON limited_users.id = bigger_user_id UNION SELECT friends.bigger_user_id AS id, friends.smaller_user_id AS friended_to FROM friends INNER JOIN limited_users ON limited_users.id = smaller_user_id }, table_name: 'users_friends'), ApplicationRecord.instaload(%Q{ SELECT follows.followable_id, follows.follower_id FROM follows INNER JOIN limited_users ON follows.follower_id = limited_users.id }, table_name: "users_follows") ], user_ids: uids, time: t) ``` the sql: ```sql WITH limited_users AS ( SELECT id FROM users WHERE users.id = ANY ($1) AND users.created_at > $2 ) SELECT row_to_json(limited_users.*) AS row, 'User' AS _klass, 'limited_users' AS _table_name FROM limited_users UNION ALL SELECT row_to_json(users_friends.*) AS row, 'User' AS _klass, 'users_friends' AS _table_name FROM ( SELECT friends.smaller_user_id AS id, friends.bigger_user_id AS friended_to FROM friends INNER JOIN limited_users ON limited_users.id = bigger_user_id UNION SELECT friends.bigger_user_id AS id, friends.smaller_user_id AS friended_to FROM friends INNER JOIN limited_users ON limited_users.id = smaller_user_id ) AS users_friends UNION ALL SELECT row_to_json(users_follows.*) AS row, 'ApplicationRecord' AS _klass, 'users_follows' AS _table_name FROM ( SELECT follows.followable_id, follows.follower_id FROM follows INNER JOIN limited_users ON follows.follower_id = limited_users.id ) AS users_follows ``` the output: ```ruby [#, #, #, ...], "users_friends"=> [#, #, #, ...], "users_follows"=> [{"followable_id"=>931, "follower_id"=>23, "follower_id"=>23, "follower_id"=>19, ...]} ```
above example with relations (no need for dynamic_attach) #get list of users, those users friends, and who those users follow, all in one request. ```ruby friend_attach = attach_on: Proc.new {|users_friend| users_friend.friended_to } output = ApplicationRecord.instaload_sql([ User.instaload(%Q{ select id from users where users.id = any (:user_ids) and users.created_at > :time }, table_name: :limited_users, relied_on: true), User.instaload(%Q{ select friends.smaller_user_id as id, friends.bigger_user_id as friended_to from friends inner join limited_users on limited_users.id = bigger_user_id union select friends.bigger_user_id as id, friends.smaller_user_id as friended_to from friends inner join limited_users ON limited_users.id = smaller_user_id }, table_name: :users_friends, base_name: :limited_users, attach_on: attach_on), ApplicationRecord.instaload(%Q{ SELECT follows.followable_id, follows.follower_id FROM follows INNER JOIN limited_users ON follows.follower_id = limited_users.id }, table_name: :users_follows, base_name: :limited_users) ], user_ids: uids, time: t) pp out['limited_users'] ``` sql output: same as example above this example printed output (same as in dynamic attach example): ```ruby #, #, #], users_follows: ["follower_id"=>3, "follower_id"=>3, ...]>, #, #, ...], users_follows: ["follower_id"=>14, "follower_id"=>14, "follower_id"=>14, ...]>, #, #, ...], users_follows: ["follower_id"=>9, "follower_id"=>9, ...]>, #, #, ...], users_follows: ["follower_id"=>19, "follower_id"=>19, "follower_id"=>19, ...]>, ```

self.instaload(sql, table_name: nil, relied_on: false, dont_return: false)

A method used to prepare data for the instaload_sql method. It returns a hash of options.

  • klass called on: if called on an abstract class (ApplicationRecord) it will return a list of hashes with the data. Otherwise returns a list of the classes records.
  • table_name: sets the name of the temporary postgresql table. This can then be used in further instaload sql snippets.
  • relied_on: will make it so other instaload sql snippets can reference this table (it makes it use posrgresql's WITH operator)
  • dont_return: when used with relied_on makes it so that this data is not returned to rails from the database.
  • base_on: for relations, a proc describing how to attach to the base table
  • base_name: for relations, the table name for the table to attach to
  • attach_on: for relations, a proc describing how this table attaches to the the base table
  • one_to_one: for relations, a boolean that flips one-to-many/one-to-one (default false)

note that the order of the instaload methods matter depending on how they reference eachother.

format data ```ruby User.instaload('SELECT id FROM users WHERE users.id = ANY (:user_ids) AND users.created_at > :time', table_name: 'limited_users', relied_on: true) #output: { table_name: "limited_users", klass: "User", sql: "\tSELECT id FROM users WHERE users.id = ANY (:user_ids) AND users.created_at > :time", relied_on: true, dont_return: false } ```

self.dynamic_attach(instaload_sql_output, base_name, attach_name, base_on: nil, attach_on: nil, one_to_one: false)

taking the output of the instaload_sql method, this method creates relations between the models.

  • base_name: the name of the table we will be attaching to
  • attach_name: the name of the table that will be attached
  • base_on: put a proc here to override the matching key for the base table. Default is, for a user and post type, {|user| user.id}
  • attach_on: put a proc here to override the matching key for the attach table. Default is, for a user and post type, {|post| post.user_id}. Returning an array from this proc will make this object attach to base records that have one of those keys. This can be used with GROUP BY and ARRAY_AGG functions to stop duplicate rows while still attaching where needed. Think of attaching a user to a post where multiple posts can have the same user.
  • one_to_one: switches between a one-to-one relationship or not
attach information for each limited_user in the instaload_sql example ```ruby ApplicationRecord.dynamic_attach(out, 'limited_users', 'users_friends', attach_on: Proc.new {|users_friend| users_friend.friended_to }) ApplicationRecord.dynamic_attach(out, 'limited_users', 'users_follows', attach_on: Proc.new {|follow| follow['follower_id'] }) pp out['limited_users'] ``` printed output: ```ruby #, #, #], users_follows: ["follower_id"=>3, "follower_id"=>3, ...]>, #, #, ...], users_follows: ["follower_id"=>14, "follower_id"=>14, "follower_id"=>14, ...]>, #, #, ...], users_follows: ["follower_id"=>9, "follower_id"=>9, ...]>, #, #, ...], users_follows: ["follower_id"=>19, "follower_id"=>19, "follower_id"=>19, ...]>, ```

Hashed Global IDS

hashed global ids look like this: "gid://meritfront/User/K9YI4K". They also have an optional tag so it can also look like "gid://meritfront/User/K9YI4K@user_image". They are based on global ids.

I have been using hgids (Hashed Global IDs) for a while now and they have some unique benefits in front-end back-end communication. This is as they:

  1. hash the id which is good practice
  2. provide a way to have tags, this is good when updating different UI elements dynamically from the backend. For instance updating the @user_image without affecting the @user_name
  3. Carry the class with them, this can allow for more abstract and efficient code, and prevents id collisions between diffrent classes.

methods from the hashid-rails gem

See the hashid-rails gem for more (https://github.com/jcypret/hashid-rails). Also note that I aliased .hashid to .hid and .find_by_hashid to .hfind

methods from this gem

  1. hgid(tag: nil) - get the hgid with optional tag. Aliased to ghid
  2. hgid_as_selector(str, attribute: 'id') - get a css selector for the hgid, good for updating the front-end (especially over cable-ready and morphdom operations)
  3. self.locate_hgid(hgid_string, with_associations: nil, returns_nil: false) - locates the database record from a hgid. Here are some examples of usage:
    • ApplicationRecord.locate_hgid(hgid) - DANGEROUS will return any object referenced by the hgid.
    • User.locate_hgid(hgid) - locates the User record but only if the hgid references a user class. Fires an error if not.
    • ApplicationRecord.locate_hgid(hgid, with_associations: [:votes]) - locates the record but only if the record's class has a :votes active record association. So for instance, you can accept only votable objects for upvote functionality. Fires an error if the hgid does not match.
    • User.locate_hgid(hgid, returns_nil: true) - locates the hgid but only if it is the user class. Returns nil if not.
  4. get_hgid_tag(hgid) - returns the tag attached to the hgid
  5. self.blind_hgid(id, tag: nil, encode: true) - creates a hgid without bringing the object down from the database. Useful with hashid-rails encode_id and decode_id methods.

Potential Issues

  • This gem was made with a postgresql database. There will be some issues when using other databases. Let me know and I will help you out how I can. The main issue as far as i can tell in the transition will be with
    1. dynamic_sql arrays. I assumed going into this that prepared statements were generated when the sql or name of the sql changed at all. This is not true, prepared statements are generated based on the underlying logical flow of the request. Because of this I was treating List types as postgresql array data types. Instead I should of converted them to ($1,$2,$3,$4,...etc) strings. This change would make them act similarly to how Lists of Lists are treated currently. This change would make necessary a major version change. There would also need to be an option to roll back to postgresql arrays on a case by case basis for backwards compatibility.
    2. instaload_sql. The instaload_sql request has hardcoded sql in it that has not been tested on other databases. This may be especially problematic due to the use of the json data-type.
  • If you return a password column (for example) as pwd, this gem will accept that. That would mean that the password could me accessed as model.pwd. This is cool - until all passwords are getting logged in production servers. So be wary of accessing, storing, and logging of sensative information. Active Record has in built solutions for this type of data, as long as you dont change the column name. This gem is a sharp knife, its very versitile, but its also, you know, sharp.

Changelog

1.1.10

  • Added functionality in headache_sql where for sql arguments that are equal, we only use one sql argument instead of repeating arguments
  • Added functionality in headache_sql for 'multi row expressions' which are inputtable as an Array of Arrays. See the upsert example in the headache_sql documentation above for more.
  • Added a warning in the README for non-postgresql databases. Contact me if you hit issues and we can work it out.

1.1.11

  • Added encode option for blind_hgid to allow creation of just a general gid

2.0.2

  • major changes to the gem
  • many methods changed names from headache... to dynamic... but I threw in some aliases so both work
  • when using dynamic_sql (headache_sql), if you select a column name that doesn't officialy exist on that model, it gets put in the new attr_accessor called dynamic. This allows for more dynamic usage of AR and avoids conflicts with its interal workings (which assume every attribute corresponds to an actual table-column).
  • dynamic_sql can be configured to return a hash rather than the current default which is a ActiveRecord::Response object, this can be configured with the DYNAMIC_SQL_RAW variable on your abstract class (usually ApplicationRecord) or per-request with the new :raw option on dynamic_sql. The hash is way better but I made it optional for backwards compat.
  • dynamic_instaload_sql is now a thing. It seems to be more efficient than preloading. See more above.
  • the output of dynamic_instaload_sql can be made more useful with dynamic_attach. See more above.
  • postgres is now a pretty hard requirement as I use its database features liberally and I am somewhat certain that other databases wont work in the exact same way

2.0.16

  • changed model.dynamic attribute to an OpenStruct class which just makes it easier to work with
  • changed dynamic_attach so that it now uses the model.dynamic attribute, instead of using singleton classes. This is better practice, and also contains all the moving parts of this gem in one place.
  • added the dynamic_print method to easier see the objects one is working with.

2.0.21

  • figured out how to add to a model's @attributes, so .dynamic OpenStruct no longer needed, no longer need dynamic_print, singletons are out aswell. unexpected columns are now usable as just regular attributes.
  • overrode inspect to show the dynamic attributes aswell, warning about passwords printed to logs etc.

2.0.24

  • added error logging in dynamic_sql method for the sql query when and if that fails. So just look at log file to see exactly what sql was running and what the args are.
  • added a dont_return option to the instaload method which works with the relied_on option to have a normal WITH statement that is not returned.

3.0.1

  • Previous versions will break when two sql attributes unexpectantly share the same value. Yeah my bad, was trying to be fancy and decrease sql argument count.
  • People using symbols as sql values (and expecting them to be turned to strings) may have breakages. (aka a sql option like "insert_list: [[:a, 1], [:b, 2]]" will break)
  • setting DYNAMIC_SQL_RAW apparently did nothing, you actually need to set DynamicRecordsMeritfront::DYNAMIC_SQL_RAW. Changed default to false, which may break things. But since things may be broken already, it seemed like a good time to do this.
  • Went to new version due to 1. a large functionality improvement, 2. the fact that previous versions are broken as explained above.
  • more on breaking error
    • got this error: ActiveRecord::StatementInvalid (PG::ProtocolViolation: ERROR: bind message supplies 3 parameters, but prepared statement "a27" requires 4)
    • this tells me that names are not actually required to be unique for prepared statement identification, which was a bad assumption on my part
    • this also tells me that uniq'ing variables to decrease the number of them was a bad idea which could cause random failures.
  • functionality improvements
    • The biggest change is that names are now optional! name_modifiers is now depreciated functionality as it serves no useful purpose. Will leave in for compatibility but take out of documentation. Used to think the name was related to prepared statements. This will lead simpler ruby code.
    • If name is left out, the name will be set to the location in your app which called the method. For example, when dynamic_sql was called from irb, the name was: "(irb):45:in 'irb_binding'". This is done using stack trace functionality. In another case the name was "app/models/report.rb:364:in 'refresh_db_methods'"
    • dynamic_instaload_sql is now just instaload_sql. dynamic_instaload_sql has been aliased.
    • Name is optional on instaload_sql aswell
    • MultiAttributeArrays (array's of arrays) which can be passed into dynamic_sql largely for inserts/upserts will now treat symbols as an attribute name. This leads to more consise sql without running into above error.
    • When dynamic_sql errors out, it now posts some helpful information to the log.
    • Added a test script. No experience testing, so its just a method you pass a model, and then it does a rollback to reverse any changes.

v3.0.6

  • Further simplifications of the library. After looking further into ActiveRecord::Response objects I realized that they respond to .map .first [3] and other Array methods. In addition to this they have the .rows and .cols methods. Feel like I should of caught this earlier, but anyway, functionaly i will be setting DYNAMIC_SQL_RAW to true by default. docs-wise I am removing any reference to the raw option and DYNAMIC_SQL_RAW. This is mainly as ActiveRecord::Response acts as an Array with more functionality.

3.0.11

  • error fix to do with multi row expressions and sql variables. No breaking changes.

3.0....

  • nil is now supported somewhat as a variable value for dynamic_sql (and instaload_sql). It seeeeems to work... i am suspicous though. The way it works is by assuming the value type boolean and having the value be null.
  • New way of creating instaload_sql, instaload, and dynamic_attach configurations which is more user friendly. Haven't put an example up yet. Good for polymorphism.
  • added some debug logs for the instaload configuration
  • changed how variables are set for ActiveRecord objects, I gave up on figuring out what ActiveRecord is doing for the most part and i now just do a eval("self.#parameter=value") type deal. Works well. Allows you to override relations when doing polymorphic stuff which is a pretty big use case.
  • I am thinking of changing how arrays are handled as that is really the only postgresql based dependency here and that will allow the library to open up to other databases. Issue is all the code I have already written in my app dependant on such things.

3.0.24

  • changed how questionable_attribute_set works again, this time by using attr_accessors on the singleton class. Seems to paper over the default reflections nicely which has been a huge issue. They use this weird delegate thing which has been throwing me off. Anyway, no more evals which is nice. This fixed an issue with dynamic attach one-to-many relation

3.1.0

  • will try keeping breaking changes out of minor version here-on-out.
  • better debugging messages and warnings for instaload
  • returning an array in the attach_on proc will now treat each item in the array as its own attach_on key instead of all as one key. So for example you can attach one user to multiple posts, and only return one user object from the database.
  • see the dynamic_attach docs above for more on that attach_on thing. ....
  • took out undocumented quick_safe_increment and safe_increment as they were not relevant to the gem - and fairly basic.

3.1.28

  • warning messages for dynamic_attach. to_s'd any provided table_names for ease-of-use.
  • logging info for dynamic_attach outside of instaload.
  • fixed an error where instaload only worked with ApplicationRecord abstract_class. Now works on anything thats a subclass of ActiveRecord::Base

3.2

  • Updates so the latest version of the pg gem and rails works with the latest version of postgresql

Questions

  • Q: does the name of a sql operation have anything to do with prepared statements?
  • A: no, the prepared statement determines uniqueness in some other way, dont worry about collisions. The only issue with prepared statements that I can determine is when you write a statement where things change every time, thus preparing potentially infinite prepared statements. This can be alleviated by using sql arguments correctly. Using arguments correctly also will stop sql injection attacks so. You know. Do it properly. Dont just hard code them into the query.
  • Q: The default name of my sql statements looks like a stack trace? Whats going on?
  • A: We set the location of where you called the function as the default name for easy debugging. Its not an error, we just take some info from the stacktrace. It also includes the method name which can provide some insite into what the query is doing. Makes logs alot nicer to look at.
  • Q: Whats MeritFront?
  • A: I am making a social media platform. Its called midflip now. See midflip.io
  • Q: Inspect on user records doesn't seem to work properly
  • A: inspect is overwritten by many diffrent libraries, in terms of devise for example, they override our override of active record's inspect. The best way to deal with this is to look at the source location of these methods and bring them together (user.method(:inspect).source_location). In my case with devise, i ended up with this in the user record:

    def inspect
        #this is the same as the dynamic records inspect execpt i also melded together some of the
        #devise stuff to keep passwords etc safe
    
        inspection = if defined?(@attributes) && @attributes
            self.attribute_names.filter_map do |name|
                if _has_attribute?(name)
                    if UNSAFE_ATTRIBUTES_FOR_SERIALIZATION.include? name.to_sym
                        "#{name}: [redacted]"
                    else
                        "#{name}: #{attribute_for_inspect(name)}"
                    end
                end
            end.join(", ")
        else
            "not initialized"
        end
    
        self.dynamic_reflections ||= []
        dyna = self.dynamic_reflections.map{|dr|
            [dr, self.method(dr.to_sym).call()]
        }.to_h
    
        if dyna.keys.any?
            "#<#{self.class} #{inspection} | #{dyna.to_s}>"
        else
            "#<#{self.class} #{inspection}>"
        end
    end
    

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/LukeClancy/dynamic-records-meritfront. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the code of conduct.

License

The gem is available as open source under the terms of the MIT License.

Code of Conduct

Everyone interacting in the ActiveRecordMeritfront project's codebases, issue trackers, chat rooms and mailing lists is expected to follow the code of conduct.