Automating Production Cost Updates: Overcoming Depth Limitations

[Gustavo_Eduardo]

Hello friends, good day. Any expert on the topic?

I have a table of inputs, another one of productions, and another one of production components.

Productions have components that can be inputs or other productions. This would generate an infinite loop if it weren’t for the fact that the same production cannot be chosen as a component or another one of later generation, but that is perfectly handled.

What I need is that when the cost of an input changes, as expected, the cost in the components also changes. I also need it to calculate the cost of the production used as a component.

For this, I have created an update column and an action to execute the update on a set of related rows which maps the values ​​of the columns… you already know this.

To avoid depending on manual action, I have created a bot to update the rows.

However, this works up to 3 levels of depth, after which, changes simply do not occur. It is evident that the bot stops working when it has already been updated once.

I need the bot(s) to run until there are no more changes in the tables. I have set the condition “Trigger another bots” but as I said, only three levels of nested productions are updated, any other solution?

[Joseph_Seddik]

I think giving the user the ability to choose a Production as a component of other Productions, is a good idea in terms of user experience. However, from my point of view, I wouldn’t just leave the relationship like this resulting in having to go into an undefined number of levels. I wouldn’t actually have an undefined parameter in the app in general.

Instead, I would have two tables: one is the current Production Components that registers what a user has selected, and another to update the Production with the final inputs, name it for example finalComponents.

For example:

1. I have production P-1 with Components as: inputs i-1, i-2 and i-3.

   • finalComponents will be the same: P-1 having i-1, i-2 and i-3

2. Another production P-2 with components: inputs i-4, i-5 and production P-1

   • Here, finalComponents of P-2 will be: inputs i-4, i-5 and i-1, i-2, i-3

Now when you need to update the cost of productions following a change in an input’s cost, your bot will work on the single-level finalComponents table, not the Production Component one.

[Gustavo_Eduardo]

Joseph, good morning, I’m going to give you access. Let’s see if you can see it. I mean, maybe you see the structure I have and from there see how it works. How about?

[Joseph_Seddik]

Con mucho gusto amigo

[Gustavo_Eduardo]

Llego a casa y te doy acceso Joseph!

[Gustavo_Eduardo]

I think it’s a good idea to limit production levels; however, I would set the limit at 15 levels to start, maybe reducing it later. I need to know how realistic these levels are. For example, in the construction industry, where there are nested productions, it’s useful; I don’t know how many levels yet, but it is. In the food industry, maybe the levels could be reduced to 5 or 6; any more would be too many. But you’re right; certainly not infinite levels or, as it’s known, “n” levels. That would be a lie and unrealistic. You’re right.

I’ve added you, Joseph, so you can see how I have the app structured and you can also make the changes. I called the table you call final components, BOM(Bill of Materials).

[Joseph_Seddik]

Well received, thank you! I’ll have a look this evening.

The setup I’m proposing involves updating the Production every time there’s a new input, so we only maintain one single level, not n, not even 2

[Gustavo_Eduardo]

Buenísimo! por si la nomenclatura de tabla te parece algo extraña, mantengo la consistencia en la app basándome en este protocolo estándar:

https://www.googlecloudcommunity.com/gc/AppSheet-Q-A/Appsheet-Work-Standard-Protocol/m-p/715222

[Joseph_Seddik]

La conozco muy buen trabajo Gus !!

[Gustavo_Eduardo]

Hello friend, could you see something? I’m trying like you told me but I’m still slow.

[Joseph_Seddik]

Hi my friend, apologies for late reply.

I’ve seen the app. You actually have multiple tables, each for one level. My proposal is to have a single table for Productions and another for Supplies.

A third table “Production_Supplies” should manage the many-to-many relationship between both tables. A part from the key, this table would have only two columns, one Ref to Supplies, and another to Productions here with Is part of option checked.

When a user creates a new Production he should be able to add supplies in the Form. Another EnumList column with base type Ref to the same Productions table, would allow the user to include to include other productions as part of this new one. Let’s call it “Included Productions”.

When the row is added and the “Production_Supplies” new entries are created, a bot will be triggered to work on the list of Productions in the “Included Productions” column. Through a recursive action, for each Included Production, the bot will copy all Supplies related to this included production in the “Production_Supplies” table and assign it as well to the new Production (if it was not already added by the user in the Form during the creation of the new Production).

At the end, for each Production, you’ll have a list of included productions as selected by the users, and also through [Related Production_Supplies], you’ll have the list of all Supplies related to the production, either directly selected by the user, or indirectly added (by the bot) from other included productions.

I hope this is clear. Otherwise, please tell me and we’ll do it step by step.

[Gustavo_Eduardo]

Hello Joseph! Nice to hear from you! Great, I’ll try what you’re suggesting! I’ll be working on it; I’ll let you know as soon as I can! Thank you very much for your response! Have an excellent day, my friend!

[Gustavo_Eduardo]

Hi friend @Joseph_Seddik I’ve tried what you suggested, but unfortunately I’m not sure how to implement it that way. I’ve created the new structure with the columns you mentioned and the relationships. I’ve added you back to the project so you can take a look if you can. Thank you very much for your help.

However, I would like to clarify that while productions can contain other productions, these can be used in different quantities. Imagine dough production, which uses flour and water. To make bread, I need dough, and to make pizza I also need dough, but in different quantities.

[Joseph_Seddik]

No problem my friend. Please tell me where you are having an obstacle.

In the third table “Production_Supplies”, you can have a Quantity column to specify the quantity of each supply in that production.

Now if you are saying that a Production can contain a certain number of another certain production, in that case you’ll have another table “Production_Productions” also managing this many-to-many relationship.

For example, let’s say Bread is a production. For this production you’ll need a quantity of water and a quantity of flour and a quantity of salt.

Your Production_Supplies table would have:

key	Production_Ref	Supply_Ref	Dose/Quantity
key_1	Bread_key	Water_key	1
key_2	Bread_key	Flour_key	1
key_3	Bread_key	Salt_key	1

Now suppose that you have another production that is DoubleCheeseburger. For this you would include three slices of the Production Bread, as well as 2 Steak, Cheese and a tomato slice.

Your Production_Productions table would have:

key	Production_Ref	Sub_Production_Ref	Dose/Quantity
key_1	DoubleCheeseburger_key	Bread_key	3

And your Production_Supplies table would have:

key	Production_Ref	Supply_Ref	Dose/Quantity
key_4	DoubleCheeseburger_key	Steak_key	2
key_5	DoubleCheeseburger_key	Cheese_key	1
key_6	DoubleCheeseburger_key	Tomato_key	1

Your bot will be triggered upon Addition of a each row in the Production_Productions table. It will then add the Supplies of the Sub_Production_Ref (Bread) column as supplies as well to the Production_Ref (DoubleCheeseburger) multiplying them by the quantity.

So at the end, the Production_Supplies table would have:

key	Production_Ref	Supply_Ref	Dose/Quantity
key_4	DoubleCheeseburger_key	Steak_key	2
key_5	DoubleCheeseburger_key	Cheese_key	1
key_6	DoubleCheeseburger_key	Tomato_key	1
key_7	DoubleCheeseburger_key	Water_key	3
key_8	DoubleCheeseburger_key	Flour_key	3
key_9	DoubleCheeseburger_key	Salt_key	3

Afterwards, if you need to include the production DoubleCheeseburger in another production Combo_Meal, you’ll have all its needed Supplies already present, all in one level.

[Gustavo_Eduardo]

Well Joseph! Thanks for your fast response! I’ll be working during the week!!! We are in contact

[Gustavo_Eduardo]

Friend. Thank you and thanks God. I have been trying to do this for more than a year and couldn’t get it done. Only God knows how much you have helped me. I pray that God blesses you, brother!

I had been approaching the problem the wrong way. As the database grew larger, it was poorly managed. It was a huge burden for the app. Now everything is resolved very smoothly without the need for major actions.

You are amazing, @Joseph_Seddik . If I could choose a ‘Best Community Colleague’, you would surely win that title.

Thank you so much. Greetings from Argentina.

I Only have one doubt left. Imagine that in the bread that has flour, water and salt (let’s say) I have forgotten to add the salt. But it turns out that I add salt to the bread. The costs are updated but the inputs are not added to each of the productions that contain this input. Do you know how I can do it?

You said “When the row is added and the “Production_Supplies” new entries are created, a bot will be triggered to work on the list of Productions in the “Included Productions” column. Through a recursive action, for each Included Production, the bot will copy all Supplies related to this included production in the “Production_Supplies” table and assign it as well to the new Production (if it was not already added by the user in the Form during the creation of the new Production).” How can I do it.

Here’s what I have in mind, though I’m not sure if it’s the correct approach or if there’s a different implementation:

If I add a delayed input to the “Productions_Supplies” table, this input will be registered with its corresponding productionID_Parent. I might need to verify if this parent production exists as a component in another production. If it does, it would return true; otherwise, false. The value of that column would be used in an iterative action, as follows:

1. First action: Add an input from the “Productions_Supplies” table to the same table.
2. Second action: Verify. If the result is true, continue; otherwise, exit the loop.
3. Third action: If continuing, execute an action on a set of rows in the “Productions_Subproductions” table and add the rows where the parent production of the component matches the ID of the parent production.
4. Fourth action: Group the previous three actions.Then implement a bot to execute the action each time a row is added to the table.

[Gustavo_Eduardo]

Friend, I hadn’t marked it as a solution since I didn’t solve everything exactly as you suggested, but I must give you 50% of the credit because it was thanks to you that I managed to think outside the box. To solve it, I had to get creative and it took me a total of two and a half years. However, from your contribution, I only took three weeks, and here I am, happy as can be.

Let me tell you about my solution.

I have a component table where inputs and/or sub-productions are loaded. Depending on the type of component, a bot is activated that triggers an action to send the elements (both inputs and sub-productions) to the appropriate tables.

The sub-productions that have their breakdown in BOM are copied to another intermediate table called Subproductions_BOM, just as they were sent to the sub-productions table.

The sub-productions that consist of other productions have a different logic since they must go through the Subproductions_BOM table to obtain the sub-productions that have a breakdown in BOM.

Every time something is added to Subproductions_BOM, a bot clones inputs in the BOM table.

If I have to add an input that didn’t exist before, an action is executed through a bot in BOM that triggers a loop, adding rows according to the number of sub-productions involved, and the loop ends when the counter reaches zero.

If I have to add a sub-production to an existing production and replicate it in all other productions containing it, I need to create an action that is executed on a set of rows from the Subproductions_BOM table. That action is precisely what propagates the inputs in BOM.

It wasn’t easy, but it’s encouraging to say that I didn’t use ChatGPT at all and I must encourage people to solve things on their own because (at least in my case) I used to rely heavily on that tool and spent more time creating a prompt than thinking about the solution.

It has been a very difficult but gratifying experience for me, and if this can be done with AppSheet, I must say that AppSheet is (if not the most) one of the most powerful platforms for creating apps. Google should consider supporting it more, and we users should promote it because it’s a versatile and democratic tool.

Friend, thank you very much in advance (I will also mark my solution).

[Joseph_Seddik]

Well done my friend, well done! Thanks much for the sweet words and sorry I haven’t seen your last reply to the post till now, but it came with big benefit

When executing recurrent actions on a list of keys, instead of using a separate counter, I just iteratively operate on the first key then remove it from this list till the list becomes empty. But it’s only a matter of preference.

As for ChatGPT, I rather use Google’s Gemini, since I’m now fully dedicated to Google Cloud’s environment. However, I never use it to give me code, I use it search for documentation, discuss best practices and architecture for multi-component solutions, anticipate risks and scalability concerns, and things like that. Otherwise, for coding the different parts, I either learn it or give it to someone else more knowledgeable about that part. But I never ask a machine to give me a code. I could copy code from repositories by people, not by a machine. Also, when I could get the advice from an expert that is always waaaay better.

[Gustavo_Eduardo]

Thank you, my friend! I’m glad you liked my message. I understand your method with the lists and what you mentioned about using Google’s Gemini. I really appreciate your approach and preference for expert advice. It’s always a pleasure to share ideas with you!

[Gustavo_Eduardo]

Regarding this, I’ve thrown myself headfirst into trying it out, and I’ve implemented this solution, and it really works, I’d even say it works better. Additionally, I’ve been optimizing the app; it’s true that I ended up with some pretty complex bots, but I think even I don’t recognize my current progress compared to two years ago. My friend, thank you so much again. I’ll never get tired of saying that you’re amazing! Thanks @Joseph_Seddik