Ask HN: Examples of reliable software you enjoy using

Inkscape: When you have to do graphics and diagrams for slides and posters, you don't have all that much choice on Linux, so I'm happy a quality vector graphics program exists. It might be a bit more cumbersome than commercial software dedicated to schemas and technical drawings, but at least you can rely on it to get the job done, with only your creativity as the limit.

Cog/Cogx, XMMS and Winamp: I don't have Windows or Mac OSX, but I have good memories of using these simple (in a positive sense) music players on occasions I had to use those operating systems, and I still miss a basic player such as XMMS these days when everything looks like it's trying to out-iTune iTunes.

Then there's mostly server-side software such as Nginx; Sqlite and Postgresql (and other database software); and typical mail server software like Postfix and Dovecot that I guess most people would agree to be vastly reliable, but that shouldn't be a surprise given the nature of server-side software.

Working with CFEngine, I got into git. I trust git too. :-)

Left to over 30 years of growth outside of its intended use case, LaTeX proper does nothing you would need it to, and the packages that you use with it are horribly fragmented. Maybe Latex is good at the things it sets out to implement, but those things aren't what I or anyone else uses it for (creating publication-quality documents). So in considering latex we are forced to also consider the environment of libraries it exists in.

LaTeX brings all the problems you had in the Windows 2000 era of programming into document creation. All your favorites like misleading compiler messages, dependency hell, choosing a compiler, knowing whether or not you have run the compiler enough times to produce a final document (!!!), bad compile environments, inconsistent naming, inconsistent syntax for passing parameters to library calls (!!!), GOTOs.

There is no way to write semantic latex for most documents where you would need it. Here's a fun project: find a way to reference the same footnote or endnote multiple times throughout a document in a way consistent with how you reference anything else (that is, using \label and \ref). Some arcane technicality in the way the counter is set up prevents you from doing this. Instead, you have to create your own global counter, write a function to maintain it, then use this function in lieu of LaTeX's implementation of footnotes.

Every time I want to do anything other than add words to my document, I have to think about whether I want to use Latex's meager facilities, facilities that I had to write, or facilities that someone else invented, then worry about all the quirks in them. If I want to add a table to my document, I need to decide whether it should be a `tabular', `tabular* ' `tabularx', `align', `array', `eqnarray', or `matrix'. Someone has told me once that one of these is very bad and should be avoided unless absolutely necessary, although he didn't tell me why, and I don't remember which one. If I want to make a cell span two columns, I have to remember whether that is the `multicols' package, or the `multicolumn' package. If my table gets too big, I now have to hard-code a split myself, or repeat the entire process and convert my table to a `longtable'. I hope I don't have to wrap anything in a `minipage'.

It baffles me that thinking about these things is prerequisite to creating a document.

The attitude that 'LaTeX is easy' is perpetrated loudly by people that haven't had to use LaTeX very much. It is only tentatively saved by the comprehensive (sometimes, if you're lucky, even readable) documentation of all its quirks.

I am not going to stop using Latex, but no one will ever get me to say that it's easy.