You are currently visiting the old customsolvers.com 2.0.
For the latest version, go to https://customsolvers.com/.

Failed attempts

List of varocarbas.com contents removed for different reasons
The R&D projects section doesn't include those which were cancelled and their contents deleted; what happened at different points and for different reasons.

The failed varocarbas.com projects are the following:
  • Project 1: Creating a proper web search engine from scratch
    . I spent on this project much more effort than in any other one. I even broke the I-write-each-single-line-of-code rule by including a WordPress blog and a small JavaScript snippet done by others. Back then, I wasn't even sure about the varocarbas.com (projects) format; that's why I created quite a few applications, collected quite a few information and wrote a lot.
    The original intention of creating a proper search-engine running on very limited hardware was quickly proven impossible. Then, I focused on building a domain-ranking system, which was also proven impossible under the intended conditions.
    The project was cancelled after the company providing the varocarbas.com hosting by then requested me to immediately stop all the applications dealing with the main data management tasks in this project. Note that, by the time all the bots were stopped, there were various databases with over 2 million records, which were being systematically accessed by a set of 20-40 applications running 24/7. Even before that, the practical impossibility of the intended goal (backlink-based domain-ranking storing a big amount of intermediate information) under the given conditions (most basic shared-hosting plan) was already quite clear. In any case, that hosting provider seemed to be managing resource over-use in a shared environment very badly.
    I kept this project for some months after its cancellation, but finally decided to delete it because of the same reason than other projects below: not exactly being what I wanted for varocarbas.com. Although this project did have various valuable outputs, it was too big, not always technical enough and too appealing to random visitors of no interest for me.
    Note that this project was the precursor of the web domain ranking which is currently running on varocarbas.com.
  • Project 2: Analysis of existing website visitor tracking software
    . I created this project right at the beginning, together with the aforementioned Project 1. I wrote the preliminary sections explaining its motivation (understanding the reasons for certain errors while associating IP addresses with geographical locations), but nothing else. I was spending all my time with Project 1 and couldn't find the moment/motivation to focus on this one.
    I cancelled it just some months after its creation and deleted it shortly afterwards.
  • Project 3: Analysis of code migration from C# to PHP
    . This project was meant to describe the C#-to-PHP migration which was required by the program described in Project 4.
    I did complete this project and even kept it during quite a few months. Finally, I decided to delete it because of not meeting my expectations for varocarbas.com. Note that I am quite experienced in both C# and PHP, so didn't really learn anything here. Most of the contents of this project were almost forced references to not-too-relevant issues. Additionally, my decision of deleting the closely-related Project 4 made it virtually useless.
  • Project 4: Innovative multivariate regression tool: trendingBot
    . As explained in the corresponding section, trendingBot has a merely historical significance. More specifically, the version being converted to PHP wasn't even the original Fortran one, but the 2.0 version which I quickly developed as a way to publish its algorithm.
    This project wasn't ever completed, but it reached a quite mature stage (i.e., online version of trendingBot up and running). I wasn't planning to spend too much time on it (or Project 3) because of assuming that a simple conversion to PHP should be enough. But this wasn't the case because the original program relied too much on memory for what a web-based version (running on very limited hardware) was able to gracefully deal with. Thus, the version resulting from a simple conversion to PHP wasn't performing too well; but I wasn't feeling like converting all the memory reliance into a DB one, much less by bearing in mind that it was a very irrelevant-to-me development. Another reason for its deletion was to minimise the surprisingly high number of visits which trendingBot is still having; even now, years after its main algorithm was made public and still more years after it stopped being actually relevant here.