what other problems do i see in linux development

The pain of not only my company, but also of my colleagues' companies, I will begin to describe with general information on the market. According to the data kindly provided by the association “Russoft” – year after year the time for developing solutions for Windows in Russia is gradually decreasing, and for Linux family systems – it is steadily increasing. In addition to the fairly obvious fact of decreasing interest in Windows in our country in principle, the increase in development time for Linux systems is also associated with another important factor. This factor is the significantly increased complexity of development, associated with the number of Linux systems, their technological diversity.

Compatibility and vendors: how to make a product compatible with all versions

According to various sources, the number of existing Russian desktop operating systems last year was between 30 and 40 (in the register of Russian software for class 2.09 – general-purpose operating systems in the summer of 2023, there were 38 different products), and this year it has already exceeded 80. We do not think that there will be a vendor that provides compatibility with all of these systems. Even with half of them. Frankly speaking, there is probably no need for this.

The three operating systems that the company has chosen and ensures compatibility with are optimal for providing a high-quality solution to the OCR import substitution problem. But even this is not easy. Judge for yourself, for just three operating systems we need to develop, test and support 7 different distributions. Customers buy operating systems, then refuse to update them. Naturally, vendors and developers always try to make the product better, not to stand still. But the customer is not always ready to update. It is quite possible that with the release of new minor, not to mention major, versions of operating systems, we will have to compile another one, or two, or three distributions. And support them. I don’t really like this situation. What are the options?

The library problem: who supports what

Why does this happen at all? In our case, it is due to different versions of libraries and technologies supported by Russian operating systems. The table shows several fairly popular ones.

Look at the difference in versions. By the way, let me remind you that not all libraries have backward compatibility. I remember more than one or two unpleasant situations when the customer nevertheless updated the OS, updated libraries that did not have backward compatibility, and the product stopped working correctly due to the absence of one important function in the old version of the library.

Let's move on. We are releasing a server version of the product, it is made by the same team that deals with the desktop. Let's assume that there will be no protected versions that require building a separate distribution. And there will not even be any request for compatibility with the fourth, fifth server OS. Even in this case, the number of distributions will grow to ten. If there are protected versions, and, for example, a fourth OS appears, then the number of distributions will grow to a theoretical 13-15. Great, right?

Let's get back to the table with the time costs for development for different operating systems. Personally, it no longer seems to me that the time for developing software for the Linux Family has increased only because of the gradual withdrawal of Windows from the market. As the organizers of the wonderful event where I voiced this problem correctly noted, only 1.5 years of truly extreme import substitution have passed, and the situation in the future may change for the better. But do you know what I'm afraid of?

Import substitution: does it help or not?

To date, active import substitution has affected mainly the public sector. Critical information infrastructure facilities, state corporations, federal executive authorities and regional executive authorities. But right now, new operating systems for the commercial sector are being developed (we are talking about the OS from Softline, where MSVSfera is taken as a basis). There are already operating systems that are not related to the Linux Family, written from scratch. And we are not talking about no-name systems (we are talking about Kaspersky OS). There is a possibility that Chinese corporations will promote their solutions in our commercial sector (EulerOS from Huawei).

Moving to the web or “clouds” for the main customers – government companies or critical infrastructure facilities – is not an option, since primarily due to the information security component, not everyone has the opportunity to use such versions of products.

The problem is broader and is now manifesting itself in the most unexpected forms and from unexpected sides. For example, here in his interview CEO of Astra Group Ilya Sivtsev talks about the landscape of the IT products market and a similar problem of diversity. In addition, I have read many publications about the problem of the number of scenarios in educational programs and IT mentoring programs. The topics are different, but the pain is similar: any unregulated entities strive to fill the entire volume and even “with a slide”. Occam's razor is needed and urgently, but – by what principle to cut? If to cut at all.

Conclusion

Colleagues, what are we going to do with all this diversity? I haven’t found a solution on my own yet, although I have some thoughts: one of the technically possible options is to reduce the technological gap between OS vendors (the gap in supported technologies is not that big); another is to move towards creating our own association of native software and framework manufacturers. We intend to follow the latter option and already have the first results (I also plan to tell you about them in due course).

I'd be glad to hear your thoughts on the above in the comments. Thank you!