Widget For Recent Posts Is Now Working

When working with an emergency project we must set priorities, testings up plugins can be exhaustive, nobody want’s to retest every location where the plugin is installed whenever and minor modification occurs, but when we are in a fast pace projects, with no planning and just one mark, proves that it worth effort, the process of testing is even worst and consuming.

So is basically working under a large growing mess of annotations, papers and folder, software for project management, but in fact nothing can really help. That is the very start moment of an project,  changes are very often and when it finally reach an first stable version, is time to get back on track and get the word left behind done.

For about months the Recent Posts Widget not working was annoying me very much, consuming and stressing, I then realized that it is almost the center of the blog, if that list was wrong, with wrong content, links, information and languages, visitors may not understand where they are, so it become a urgency.

Now I can share all my blogs links, without getting people confusing asking themselves why that Recent Posts links is showing that weird stuff, now everyone can navigate safe and be related with the right content in the right time.

 

A huge problem emerged just a minute before our costumers tests our first WooCommerce virtual store

When we just finished up our local tests of our first online WooCommerce store, when it was online and ready for our first costumers to test it out a huge problem emerged. When we use native WordPress functions and the WooCoommerce plugin for quick attach an virtual store to our wp site, we must adapt the data managements to a mix of combined patterns of both, wp and woo, they are all good, but have problems for scalability.

WooCoommerce stores in the same table products and orders (side to side with WordPress blog posts, pages, menus…) it is quickly for deploys new stores, but in few moments everything can become a mess of data, stored mixed. The problem is when we have to update only certain product data, like stock quantity, and of course keep orders preserved, it is impossible to do (without temporary lock down live site for maintence) with bult-in WordPress, WooCommerce or even SQL commands, it is basic a lock-up.

To break that, the solution we take is a dirty little hack, that is only acceptable by expert users, who fully understand what we suggested and coded as a solution. We just released or fist hack plugin, it is avaiable on github and soons as we make a good documentation we will submit to wp plugins directory:

What F5 Sites | Woocommerce shop_order Tables does?

It creates a hack table on database called 9woo + wp prefix + shop_order + endpoint (posts or postmeta), and (by using hooks) whenever WordPress or WooCommerce asks for orders data it force $wpdb object to get if from another tables. It is a hack, you must have a deep understand about hook and wp core functions before you can take the most of our solution.

Our first public and opensource WordPress plugin

plugin: f5sites-shared-post-tables-and-uploads-folders

After almost 9 months of tests and improvements, a lot of risk taken, we got in alpha stage of development for Shared WordPress Post Type and Taxonomies + Uploads Folder. The projects uncertainty itself seems not enough, if it was about risk taken, that new plugin initiative was a big all in, because I it stopped every single project for more 6-9 months, again, latening market entrance.

Years ago, 2014, a recent bachelor programmer said to me that WordPress data model was a mess, I took it personally and go look for more informations and sadly find out that he was partially correct, the simplicity of it’s model was very good for some situation, but locks out the developer in some scalability problems, if not well configured.

WordPress development is already fast, it has dozen of bult-in functions and integration, for start a new project is easy as use Fantastic or other script installer to get site ready in seconds. But for quickest deployment ever it was a prision, one installation for each project is the start of a big mess.

So the first step of rearrangement was simple using one single install for all blogs, and take the code dependency management to the next level, using Composer. Next is share all the users across single installations, it was a huge progress, and finally why don’t share blog posts too? In that point we can start answering where is the “mess”.

Objects are stored mixed in database schema, blog posts is stored as the same global type of products, pages, menus, contact forms, and every object called post_type. To make things more complicated there is a separated table for postmeta, it is a relational table with same index (post ID), so you must let wp address it’s number, and using a second table will make unique identifiers desync, it is very hard schema to break.

But came to me that the solution was easy, don’t take me too long to realize that it is actually better than worst, I can now share pages, menus, contact forms, midia, products, everything, withou losing control of some data stored in single databases, for each single installation. It was just a question to make it works well, sharing post_types and don’t break-up WordPress, all of it without touching core code, just by enabling/disabling a plugin, and after 9 months of intensive tests and twice almost abandoned it, finally it get stabilized and ready for refinements.

For separate singe install shared data I used a category, and for automatized more the process the category is exactly the domain name, so just check the domains when creating a blog post, and will be displayed only in that domain. For products it works even better, it list all products and just indicates and redirects the user to the correct store domain, it was great.

It also inaugurate a new era of development, focusing most effort in making public open-source software, hosted primary on open Github configuration, for free download and re-utilization, we perceive a new paradigm of value on it initiative. It stills in alpha, but already available for public.