F411 USB

March 7, 2025

Black Pill boards - F411 USB -- Cscope

There is a parable about two men having a contest to see who could cut down a tree the fastest. One got right to work with an axe. The other sat down and spent some time sharpening his saw. The one with the axe called the other lazy and such, but then the man with the sharp saw got to work and quickly finished before the hard working man with the axe.

New and sharper tools are always a question for me. Is it worth the time to learn a new tool, or should I just get busy with the tools I already use every day and am familiar with. Sometimes there is no way to predict. You just have to try new things. And sometimes new things don't seem better right away, it takes some time to get familiar with them and learn to use them effectively.

So, I want to add Cscope to my "bag of tricks". Just for the record, there are two other tools you might consider:

Sourcetrail - see my notes below. This is an abandoned project. The source is on github, but I haven't tried building or using it.
Codequery - see notes below. This seems to be an active project, but I would have to build it myself (it is not available as a handy package for Fedora like Cscope).

Cscope for me.

Cscope on Fedora

I currently us ctags regularly and find it invaluable. I use vim, and have configured vim so that certain keys trigger the most important ctags actions. What ctags does is to find the definition of things for me. Usually the "thing" is a function. I see a call to a function, and by typing Crtl-] I get taken to the function definition. The Ctrl-T key takes me back.

What I wish for (and what I ended up with if you persevere to the end of what is now this 3 part series) is that I can type Ctrl-\ and get the inverse lookup via Cscope. Cscope is capable of more, but adding just this one thing to Vim satisfies my desires.

What I am missing is the reverse. I cannot sit on a function definition and ask the question - "where is this called from". Cscope offers to give me that.

Just for the record, I am using the current edition of Fedora linux (Fedora 41). To generate the ctags "database" I just type "ctags -r ." at the root of a project and I am ready to go. There are at least 3 different versions of ctags and I have no idea why there are 3 and what the differences are. I once typed "ctags --version" and it told me it was "Exuberant ctags", but today I typed that and got:

Universal Ctags 6.0.0, Copyright (C) 2015-2022 Universal Ctags Team
Universal Ctags is derived from Exuberant Ctags.

So, Fedora moved from exuberant to universal. I never knew, and clearly it works just the same (from my point of view), so I don't care or mind.

Install and setup Cscope

First I have to install it via:

dnf install cscope

I just try to run it. I starts, but it is not clear how to exit the program. It has remapped Ctrl-C to toggle what it calls "caseless mode". I would like it to use my vim habits (j and k to move up and down), but it wants me to use arrow keys. Perhaps this can be configured? It offers to "change strings", which worries me. I want this to be a read only tool with no ability for me to make mistakes and change anything.
So some learning is required up-front. I am going to start a new page with my own "tutorial" incuding worthwhile links.

You type Ctrl-D to exit.

What about CodeQuery?

This is not available as a package for Fedora that I can simply install. I will need to clone the source and build it. This is something I can certainly do, but I'm not going to do it right now.

The description of what codequery does seems rather "exuberant". They have a long list of languages they hope to support, which worries me. Perhaps it shouldn't, but they seem to be taking a big bite. Someday I will have to try it. If I wait long enough, maybe it will make it into the Fedora packages.

Sourcetrail

There is yet another project "SourceTrail". It was developed at Google (under the name Coati) starting in 2016 and made open source in 2019. The project was abandoned around 2021.

The notice says "we lost interest in language analysis". They also mention that the software was "ambitious", supporting multiple languages and platforms. They found it to be "a lot of work that is often simply not fun".

All of that is easy to understand. We see this all the time on Youtube channels, where people burn out, often because they commit themselves to doing too much. I tell people that if I am not having fun, somebody needs to be paying me -- and even then the best answer is often: "no!".

Feedback? Questions? Drop me a line!

Tom's Computer Info / tom@mmto.org