LaTeX is becoming a mess

As almost any working mathematician, I am a TeX user. I began with (AMS-)LaTeX in 1998, when I was writing my graduation thesis. Some years later I learned the fundamentals of plain TeX, because I wanted to have a deeper look at LaTeX's philosophy. Of course I always use LaTeX for my preprints, since plain TeX is really too... plain for contemporary publishing.

While plain TeX was frozen by its developer Donald Knuth, and the compiler will not be modified any longer, the developers of LaTeX are rather active. Apart from the mysterious LaTeX3 project, new tools are added to the LaTeX system. In the last months I have been fighting against a few recent changes that made my life more complicated than it should have been. The first "criminal" package is biblatex. From its website,

Biblatex is a complete reimplementation of the bibliographic facilities provided by LaTeX in conjunction with BibTeX. It redesigns the way in which LaTeX interacts with BibTeX at a fairly fundamental level. With biblatex, BibTeX is only used (if it is used at all) to sort the bibliography and to generate labels. Formatting of the bibliography is entirely controlled by TeX macros (the BibTeX-based mechanism embeds some parts of formatting in the BibTeX style file. Good working knowledge in LaTeX should be sufficient to design new bibliography and citation styles; nothing related to BibTeX’s language is needed.

Well, it looks interesting, doesn't it? Since I like the classicthesis style for my preprints, and since the sample file points to the biblatex package, I decided to keep it in use. The learning curve is not steep, and in a few hours I was able to master the differences between the old bibtex and the new tool. I should confess that mathematicians and mathematics journals do not use strangely formatted bibliographies like they do in human sciences, but biblatex is described as the future; so, let's use it!
As I said, any typesetter can easily move to the new syntax; the troubles begin when you decide to put your manuscript on public databases. One major example is the popular arxiv.org archive.
I began a new submission, I uploaded the tex source and the bbl file (as I did with bibtex), and... oops! The submission was rejected due to a fatal error in the compilation. The error was really impossible to amend, since exactly the same file is compiled on my computers with no error at all. From the logs I could see that arXiv leans on the old TeXlive 2011, while I use the new TeXlive 2012, and the biblatex package has been updated several times in the last months. So, no hope to place a preprint on arXiv, if you use biblatex.
The only way out was to move back to an old-fashioned bibliography. Since bibtex works well in TeXlive2011, it was the natural choice, and finally my preprint was accepted by the arXiv engine.

Then came the submission to a journal. In the first step, the author sends a PDF file to the editorial boards, so everything goes smoothly. Once the manuscript is accepted for publication, the author is kindkly requested to send the source file, and often the source file should comply with the journal cls style. And this may be a nightmare: almost no journal currently accepts biblatex, and a lot of them wants a static bibliography (\begin{thebibliography}...\end{thebibliography}) at the end of the source file. Now, even if you use BibTeX, the effort is very small: just copy the content of the .bbl file into the source file, and you are done.
But if you are a biblatex user, you might just as well waste an afternoon on the internet to find out how to recover the static bibliography. And the answer is: you can't! Indeed, the .bbl file created by biblatex has a completely different structure, and must be processed again by the pdflatex compiler. When I tried to explain this to the editor, I got a sad reply: "I do not understand what I should do: I click on the "build" button on my computer, and there is an error. I think you must solve this problem, since you sent me the file. Best regards."
And now? Well, I move back again to bibtex, created the standard .bbl file, pasted it into the source, and sent everything to the editor. He/she was really happy to click on his/her button and see that a pdf file was produced!
I learned one thing from this experience: always be conservative!

But I could mention more examples. Since PDF files were introduced as the best format for web publishing, the LaTeX developers moved from the latex+dvi engine to the pdflatex engine, and the .dvi files were doomed to extinction. But some publishers still try to compile the source file with latex, and they get errors if the authors uses packages developed for pdflatex only. This happens with non-standard fonts, with XeLaTeX files, with pictures and images.

In my humble opinion, LaTeX is undergoing an evolution process and will probably lose its most important feature: once upon a time, LaTeX was a universal engine, now it works only if you and your collaborators use the same release, with the same fonts, with the same upgrades. And, sometimes, with the same editor and with the same character encoding!

Edit: yes, of course it is my fault. I like "bleeding edge" software, I use Archlinux which may become totally unusable after upgrading, say, Firefox. I understand that a journal cannot wake up one morning and discover that the whole production is frozen because the new biblatex package has a bug.