Dear Matthias! *> What is the modularization project?* The Sage developer community has long been aware of the severe problems that the monolithic design of Sage has brought. See in particular the lively 2016 sage-devel thread "How we develop Sage" (https://groups.google.com/g/sage-devel/c/29ndCD8z94k) initiated by William.
I looked at a few of the messages in this thread (there are something like hundred messages, I feel unable to read them all). I would like to make two remarks * 26 persons participated in this thread, roughly 10 of these are still active * many of the messages were quite critical regarding modularization of the math library in sage * it seems to me that at least a big part of the thread is about modularization of the *infrastructure* of sage, not the math library. In any case, I do not see consensus in this thread. Of course, there may be other threads that I am not aware of. I was not there at the time, and only very recently realized some potential consequences of the direction the modularization might take as indicated in https://github.com/sagemath/sage/pull/36964, and formulated some questions there. To my surprise, at least Volker issued similar concerns in the thread above. I could not find satisfying answers to his questions either. Also, I must say, that I find "Don't fear. This has been thought through carefully, and you can read about it in the Sage developer's guide." in place of an answer on the edge of abuse. It is at least condescending. For convenience of discussion, I am reproducing my questions below. Martin 1.) Is there an example for someone who did not want to use sage because of some dependency of the math library? Or at least a possible reason? @kwankyu's comment above suggests that having something in the "wrong" distribution wouldn't be a big deal. But this begs the question: who profits from cutting the math library into pieces (which look very arbitrary to me and have a curious emphasis on discrete math topics)? My fear would be that at some point there is a request not to use symbolics in some module, because Lisp is hard to install on some system. (I don't think this fear is unjustified: in the section of the developer guide you pointed to, I find > The imports of the symbolic functions ceil() and floor() can likely be replaced by the artithmetic functions [integer_floor()](https://doc.sagemath.org/html/en/reference/rings_standard/sage/arith/misc.html#sage.arith.misc.integer_floor) and [integer_ceil()](https://doc.sagemath.org/html/en/reference/rings_standard/sage/arith/misc.html#sage.arith.misc.integer_ceil). OK, so some user of that module happily replaces the two functions. Now, I come along and would like to replace some other implementation by a call to something defined in symbolics. But that would be breaking a promise to the user, so it would be really hard to justify. In fact, this happened to me already, in some sense. I noticed a function definition in `sage.modular.multiple_zeta` with misleading documentation, which could be replaced by a call to code in `sage.combinat`. However, this is *already* hard to do, because it might affect performance (which is a very valid point in my opinion). I think it would be extremely bad to make it even harder. 2.) If this is about dependencies on other software, why aren't the distributions named after these dependencies? (Of course, at some point dependencies might change, for example, there might be a switch from glpk to scip.) Before I read - by chance - `distribution = sagemath-graphs` somewhere, I thought one would "modularize" things like the repl, user interfaces, and perhaps some low level stuff. But it seems to me now that this is really about the modularization of the mathematics. Also, I find it hard to believe that it is about dependencies, because the stuff in `abstract_tree.py` and friends has no dependencies on external software (unless you want to LaTeX them, perhaps). -- 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/7aafafbc-7992-4bfe-abcd-f4b433b29c3en%40googlegroups.com.
