On Tue, Jan 31, 2023 at 11:19:02PM -0300, Leonardo Brás wrote: > Hello Paul, > > On Tue, 2023-01-31 at 13:45 -0800, Paul E. McKenney wrote: > > On Tue, Jan 31, 2023 at 06:09:50PM -0300, Leonardo Brás wrote: > > > Hello Paul, > > > > > > I am a newbie Kernel Developer, and I was already a fan of your work on memory- > > > barriers.txt. > > > > > > A couple weeks ago a friend of mine recommended reading perfbook, and sent me an > > > older version (2011.01.02a). After reading a few pages I went on and found that > > > the book had newer releases and is often updated and maintained with git, which > > > got me really impressed. > > > > > > (I had a lot of work gone to waste on other books I found errors on, and > > > perfbook's git workflow just make a lot of sense to me) > > > > > > Since it's a topic I am really interested in, and I noticed the book has updates > > > from a couple days ago, I decided to install texinfo packages and compile a new > > > version for me to start reading. > > > > > > I then noticed it would be nice to keep reading the latest version whenever an > > > update came by, so I created a gitlab-ci to re-generate the book after a new > > > commit / push: https://gitlab.com/linux-kernel/perfbook > > > > > > Is that ok if I keep this gitlab repository re-generating the book? > > > > > > Is a gitlab-ci of any use for you? > > > If so, I can send it as a patch (a single .gitlab-ci.yaml file), and refine it > > > according to your needs. > > > > > > Thanks for keeping the book updated! > > > > Hello, Leonardo, and glad you like the book! > > > > On the automated CI, I cant't think of a problems with it. In fact, in > > my experience, automated testing is usually a good thing. I am adding > > the perfbook email list in case others have advice or other guidance. > > > > Just out of curiosty, are you thinking of automating tests of the > > sample code as well as of the LaTeX builds? > > For now it only automates the LaTeX build of perfbook.pdf, perfbook-1c.pdf and > perfbook-eb.pdf, but there should be no problem automating tests too. (I mean, > once I understand how to run tests.) Good point, building the tests should be a matter of "cd CodeSamples; make", but yes, the actual running of the tests is at the moment a matter of a bunch of random .sh scripts. I will think about the feasibility or not of adding something like "make check" to those directories. Thanx, Paul
