On Friday, December 17, 2021 at 2:33:55 PM UTC-8 Randall wrote:
> Hi all: > > I am building sage version 9.4 from source on my OpenSuse Linux Leap 15.2 > OS and like the install document stated, it has taken hours (using a HP > Z420 workstation with 6 Xeon processors) > > I have noticed some odd things occurring which caught my attention as at > one time in my life I was in charge of 1+million line code handoff to > another company and our code had to be very high quality. > > Here's what caught my eye: > > 1. dozens of warnings during Cythonizing of comparing signed variables to > unsigned. This warning needs to be heeded and the code fixed > 2. dozens of warnings of compares using uninitialized variables. It goes > without saying that using uninitialized variables is only asking for > trouble later on. These need to be fixed also. > 3. hundreds of warnings during package compilation, the source code does > need improvement. I did notice that some C programmers were attempting to > write C++ code, but this is asking for trouble too. C++ is not advanced C, > despite the popular (mis) belief in this. > > I am attempting to install the optional packages just to see what happens, > for the build process, but it looks like bundling several packages (say x) > on the ./compile line is only asking for repeats of the whole compile run > x-times. It should be possible to install all on only one run, not x-runs. > Do you mean the "./configure" line? If you do "./configure --enable-qhull --enable-libsemigroups" at the start, then "make" should just increase the build time by whatever it takes for those two packages — it shouldn't repeat anything. I have some questions here > > 1. Will "make all" build all the optional packages and avoid the > experimental? > (I did NOT try make all, because just doing make itself chewed up about 9 > hours time) > No, "make all" just builds the standard packages and the documentation. 9 hours seems way too long. Are you building in parallel (step 7 in https://doc.sagemath.org/html/en/installation/source.html#step-by-step-installation-procedure)? That should speed things up. For instance it takes under an hour on my 2017 iMac, not a super-fast machine, although it does use some system packages to avoid building time-consuming packages like gcc/gfortran. > > 2. I noticed that sage recommended several OpenSuse system packages on the > zypper install line, but I did run this and believe it or not, nothing was > available. Is there some consideration to creating a public mathematical > software repository where these OpenSuse packages are available? (This > probably needs to be extended to most of the public Linux systems) > We rely on package managers for the different linux systems to produce packages and then we use what's available. (Some Sage developers help with those packages.) I don't know how receptive OpenSuse would be to a repository of packages that they hadn't vetted; would they welcome this or view it suspiciously. For the way we use packages now, see https://doc.sagemath.org/html/en/developer/packaging.html#allowing-for-the-use-of-system-packages and for example build/pkgs/arb/distros/opensuse.txt. > 3. Can the dochtml installer be decoupled from the regular install, and it > run optionally? It takes up a lot of time for it to discover that > everything has already been installed. > Running "make build" will build the standard packages and not build the documentation. -- John -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/ac46712f-4fa4-42fe-b972-f8fa70fa6b6dn%40googlegroups.com.
