Closed dimpase closed 8 years ago
You don't have the git repository, this program is meant to be run directly from the repo.
Do you mean make
, or some other program? The 1st line of Makefile says
export REPO_ROOT:=$(shell git rev-parse --show-toplevel)
needless to say, if you follow your TL;DR in Readme, that's the error you get, full stop.
The readme starts with "git clone https://github.com/sagemath/binary-pkg.git"; If you do that then you have the git repo and everything will work fine. Which steps did you do that did not work for you?
On Sat, Mar 26, 2016 at 11:33 AM, Volker Braun notifications@github.com wrote:
The readme starts with "git clone https://github.com/sagemath/binary-pkg.git"; If you do that then you have the git repo and everything will work fine. Which steps did you do that did not work for you?
hmm, I probably did clone it into SAGEROOT... The readme says nothing about where it has to be (one more reason to have it as a git submodule of Sage)
— You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub https://github.com/sagemath/binary-pkg/issues/6#issuecomment-201775424
You can clone it wherever you want. By design you you must compile Sage in a special directory whose name is decided by binary-pkg so it makes no sense to have it located inside the Sage repo.
It is ludicrous, and a clear regression, that you cannot make a binary release of your custom Sage tree. It didn't cross my mind that it could be the case.
On Sat, Mar 26, 2016 at 10:23 PM, Volker Braun notifications@github.com wrote:
You can clone it wherever you want. By design you you must compile Sage in a special directory whose name is decided by binary-pkg so it makes no sense to have it located inside the Sage repo.
— You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub https://github.com/sagemath/binary-pkg/issues/6#issuecomment-201942052
Nobody else creates binaries from random builds, and for a good reason.
Besides, what we did for years didn't and couldn't work consistently. It worked after fudging around with library versions for the most popular distros but never had a chance of being a 100% reliable solution. Not to even mention the security issues.
I am not saying we need to resurrect the old unreliable setup with LD_LIBRARY_PATH, but "noone else creates binaries from random builds" sounds very hard to believe. Besides, what does prevent binary-pkg from being adapted to making binaries from a random build, besides few minor tweaks?
Fundamentally, its easy to truncate a null-terminated string but you can't grow it without deep understanding of the binary format. Hence if you bake in paths that are too short then you can't change them.
It explains why one needs to rebuild everything, but it does not explain why there is no option to rebuild an existing source tree, in a custom location with long enough paths...
On Sun, Mar 27, 2016 at 8:28 AM, Volker Braun notifications@github.com wrote:
Fundamentally, its easy to truncate a null-terminated string but you can't grow it without deep understanding of the binary format. Hence if you bake in paths that are too short then you can't change them.
— You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub https://github.com/sagemath/binary-pkg/issues/6#issuecomment-202007501
it is totally unclear from docs or sources how the thing is supposed to know which Sage root it should take to build the distribution.