Ranjit Patil

Hi, my name is Ranjit Grou. So, basically, I belong to Pune, and Maharashtra. So, I have good command over, like, PHP core, PHP Laravel. MySQL, WordPress. WooCommerce. So, I have experience of, like, 8.5 years in this skill set. I was working with X1 Solutions in the recent organization since last, 4 years, from 2020 till 2024, 28th June. So, basically, I have, also, mostly worked on, open AI development where, like, we have trained the model via API and then, we have done fine-tuning. We have directly sent the prompt to the model and then, we received the response. So, we have trained the model in such a way that whatever the request or prompt has been fitted to the model, we'll be returning the JSON format, and that JSON format we have encoded within the short-handled code understanding. And based on that, we have created multiple kinds of affiliate programs and all. So, where, like, product suggestions, and, like, applying discounts to the people if they are searching simultaneously for the same thing, then we have enabled one more gateway there. They can pay half payment and then part payment needs to be done after a month. So, this kind of addition functionality, we have enabled via open AI. Recently, I was working with Lee, Laravel for creating a SaaS. So, basically, we created QuickBooks Io, which was an open platform, has been created where, like, if any full commerce or Voodoo or, we can say, like, Shopify will be getting created, any orders will be directly sent to QuickBooks. So, this kind of a feature we have developed and based on the entity consumption, for like, if the customer is using, like, 1, 000 APIs in a month, so, based on the first 100 APIs, we have sent them for free. And, onwards, like, we have charged each API point at $40.40 per each kind of a transaction, entity transaction. So, it will be creating multiple billing platforms where, like, we have enabled and switched the panel as well as we have enabled the Razorpay gateway over there, making the subscription as well.

So, back-end PHP service, I can say, for instance, there might be several factors where the web service is in the old version, which needs to be maintained in a very structured way or where a very principal prototyping followed while creating those services. So, basically, design pattern is nothing but the normalization and the pattern which needs to be very clear and crystal clear and easy to understand for other people to contribute on that. So, according to me, that is the design pattern or design principle I have. So, basically, while creating kind of additional lengthy functionality or very complex functionality, we have to follow the object-oriented programming where namespaces, classes, traits have to be used, and auto-load has to be used for if we are using multiple libraries. So, this kind of set parameters are there, as well as we have tools to provide the commands, like rate limiter and all, which automatically enables the usage of that particular API or the particular services. Along with this, I can say, authentication and authorization needs to be managed in a very appropriate way. For instance, if an admin is getting logged in and then he does not want to have all the control, so he can delegate few of the models to a super admin. So, basically, this kind of authorization needs to be done. And for maintaining this kind of authorization, we need to have very certain coding standards where PSR coding standards needs to be followed in a very certain manner. So, cookies and cache need to be stored in such a way, for instance, that has to be mapped with a user or specific model as well. For instance, if for login we are creating multiple sessions and cookies, so within login, we have to pass an array for each user, and each user will be having that certain set of parameters value into a single session. So, this kind of a certain design prototyping of coding standards needs to be maintained to have a flexible flexibility to others as well to contribute on the same.

Yeah. So in PHP within PHP, there are certain kinds of debugging procedures. For instance, we have custom error handling or exceptional handling, like try-catch blocks out there, as well as if we want to debug any kind of bottlenecks or for instance, there will be a max execution time or error for single functionality, which goes into a loop and loop. So it will automatically stop the server or it will automatically stop the page from serving. So, basically, in this kind of feature, in this kind of scenario, we can have the debugging with print requests. So based on print requests, we can identify whatever the breakpoints are there and those breakpoints are unable or not. So based on this code execution, we can have this kind of certain aspect as well as control or the code as well. So this has to be used. But, in my experience, from the last 8.5 years, exception handling has to play a very good role in this. So until we are not providing the exception handling, the code will not work in the expected manner we were expecting. For instance, recently, I have worked on a client where he was using the Zoho integration API. So, basically, they were sending one request to Zoho, and then based on the Zoho response, they were operating other parameters or other models. So, functional making this functional based on exceptional handling is very effective. So if the response is getting empty, there's no way to point out the next module to be run, because that model doesn't have anything to perform actions and all. So this kind of additional exceptional handling needs to be done very effectively, and logs have to be maintained. So whenever we are creating any kind of functionality, if the functionality fails, a log has to be maintained in certain files or somewhere, so we can track it while debugging and fixing the code.

Yeah. So, basically, there are multiple ways to optimize the SQL query. From my side, I think, like, there was the table which had around, like, lacks of records. So, a select query, obviously, that won't work. So, there are multiple ways to function this. So, basically, the first way, which is very popular, is to have indexing on the particular table. So, indexing has 2 stages as well, within the QD indexing or the table indexing. So, basically, if we are indexing the table, that will be the permanent solution, whenever we create or insert any records, so it will be automatically considered at the indexing column and all, whatever we have mentioned. And, like, on the basis of QD execution, we have also the option to provide indexing for a particular column based on the QD execution. So, for an instance, like, 6 stars from the table name, and then we can index the column. So, that is more of an operational way. For an instance, like, one table does not always have a single entity to be indexed. So, the next time, the same table might have the other column as an index. So, the temporary indexing, which can be done via SQL query, is more relevant. Also, like, normalizing the table. So, normalizing the table can be defined as a very pro manner, because, like, if the table is very normalized, the query obviously won't get conflicted. We need to avoid writing subqueries until it's required. For instance, like, if we are joining a few kinds of tables and then we are returning single records. So, for instance, there are 3 tables like employee salary and then employee details. So, this has to be into a very normalized way. So, when we write the query, it will be performed very fast if it's normalized. If it's not normalized or not organized, to write the query, and then again, it will be getting more complicated.

So for instance, recently I have learned the materialized design long back, like, 4 years ago. So at that point, I have seen the aspect or the usage of SCSS, which is highly scaled. So, another example is CSS, which has multiple lines and is very refactorable. That has to be converted into SCSS, which turns really fast as well as that has to be done in a segregated way. For instance, a single file is there, which has 2 to 4 UI models. That is not required yet, because the first window or the first model on the single screen will be served. Our single tab will serve the single model until we are not customizing or not having an impact within the models. So, whenever the single model is getting served, that particular CSS has to be separated. Like, model 1 has to be model 1 SCSS and model 2 will be having the model 2 SCSS. Will it be very dedicated? And the common functions or the common classes or ID CSS has to be moved to global CSS. So this can be a good approach, because I have used this for SaaS as well. Because SaaS were having around 70 MB directory. So in the case where around 175 models had been created, and I have created the separate model, submodel, parent model, then based on this direct structure, I have created the CSS, and then I have called them based on the model names and all. So segregation of the CSS has to be the very first approach while having large CSS code based optimization. And second will be the SCSS to CSS conversion or single line CSS conversion or minified CSS conversion needs to be done.

Yeah. So I think the core PHP or the very vanilla PHP application, we'll be having a time load of 7 to 8 seconds. And when we move to Laravel, it will be shortened to 2 seconds. So, obviously, Laravel will have a very structural manner or pattern of code declaration and all. So, there are certain steps. So, basically, according to me, first, we have to convert all the UI within the model, and then each model has to be called on different models based on their relational relation. And, as well as, we have to keep in mind that what are the microservices or auth-related APIs are there, that have to be added in the route. So, first, we have to create a model and then the approach will be creating a controller. And, then, the last one will be creating APIs within the route. So, this can be done, but this is the kind of very hectic way because it might require migration. It might require Cedar. It might require the design common libraries and all. So yeah. I mean, that can be done, but in a very sorted manner, we have to keep in mind, model and then what are the APIs it has to be registered. So, this can be done within this pattern. Also, I would like to add one more thing here. Like, recently, I have done that. So, one of the plugins, like QuickBooks connector, that we have converted to one of the Laravel SaaS templates. So, while converting, we found that multiple user relation tables have been already there. So, those entries have to be considered while migration and all. So Cedar has to be maintained in a certain way. So, yes. I mean, by performing multiple principles, that can be achievable in very short time as well.

Yeah, so I think the user is one of the models which has been created while creating a migration and all. So the user variable in the first line, I can see the user variable will be having all the users' details. All the user rules will be considered. And based on the user rules, the forex loop has been created and then the user will load post. So what are the posts have been created for the user rule that has to be shown or that has to be bound very well. So in this, I think, like, has a property, like, have a property, some call it a property that has to be maintaining in a relationship manager. So the relationship manager, I cannot say relationship manager because it has a property because post is already one of the table. So that table has to be mapped with users. So basically, each post table, each row will be having the user ID. And based on that user ID, it has to be mapped whether the user is within the loop. So basically, the user has post can be a solution on this.

So singleton application or singleton function, or singleton pattern can be considered as the instance has to be declared or initialized or created only within the class or the ones we can say. For an instance, the code which has been given will be having a private function constructor. So, basically, the constructor will be preventing the other instances. So, basically, for an instance, whatever the constructor contains, like initialization of the variables and all if the entire directory is used with multiple variables. So that variable has to be maintained in the private constructor. But, that should not be or even the singleton instance.

Yeah. So, microservices in Laravel for ecommerce, I can see there will be certain parameters called products and then product relationship schema has to be designed. And then product will be having the categories and subcategories and variations, and each variation will be having the attributes and all. So, for an instance, we can consider WooCommerce where multiple APIs have been provided by WooCommerce. Considering multiple APIs of WooCommerce will be used for Laravel, like Laravel design and designing all the structure architecture. So, we can create multiple models for user and product. For product, it will be having multiple screens, like the cart, single product page, checkout, order, and then shop page as well. Within the single model, multiple sub models will be created, and sub models will be having the data by using the microservices. So it will be, for instance, we have created a microservice for cart. We have created a microservice for checkout. We have created a microservice for order, by adding the middleware of user or auth, we can say, for a particular logged-in user only that these microservices need to be served. And for shop and all, or for a single product page, microservices I don't think microservices need middleware, like auth and auth. Mostly, it will be returning all the data.

Yeah. So real-time synchronization, I can say, like, with the Ajax as well, we can do, basically, for instance, consider a search bar where multiple 1,000 products have been displayed on a single page. And then, at the admin side, the admin is adding products simultaneously. So that has to be reflected on the frontend. So whenever the admin creates a product, that has to be set to SQL. And then, from the SQL, each time that data has to be fetched very frequently. So without WebSocket, I can say, like, we can add timers or something called a cron job. So, basically, on the basis of a cron job with a 1-minute interval, the cron job will go to the database, read the record count, and the existing count of the loaded side. We can consider, for instance, one screen is there. On that screen, we have calculated the count of current products and set it to session. So here, on AJAX or on the cron job, we'll be getting the count of session and count of SQL queries. So if the SQL has 1,050 products and the session has 1,000 products, the 50 products are the difference, and then it will be counting the 50 products more. And then it will be syncing to session and all, then that can be done with AJAX.