Arch Linux

gmp-ecm is a well-known implementation of the Elliptic Curve Method (ECM) for factoring integers. It is a dependency of sage-mathematics.

Maxime, I see the ecm package is not used by anything else. Could you remind me what it does?

ECM removes (while UnECM restores) sync, ECC and EDC from CD ISO sectors, which are used for error detection. Since compression algorithms handle them quite poorly and already have error detection mechanisms anyway, the idea is to make ISOs much more compressible. ECM is mainly used for archiving PSX ISOs.

I could rename the package ecm-tools (windows binaries tend to be distributed under that name) if you think sage's ecm should be named just ecm. Not quite sure what to call the binaries though, off the top of my head I'd say bin2ecm and ecm2bin are likely candidates. However unecm is a symlink to ecm; when invoked as unecm, it is equivalent to ecm d, so I'd have to patch that in. What do you think?

Sage would also require patching if the gmp ecm binary is renamed, but nothing too bad (there's just one call that I know of). So I'm fine with whatever you guys decide. If gmp-ecm is renamed, please push it to staging first so I can fix Sage.

I created a fork on GitHub (since ECM's upstream is pretty much dead) named ecm-tools and pushed a package with the same name in [community], replacing ecm. The package now provides bin2ecm and ecm2bin, no more conflict so I'll close the task.

Antonio, can you wait a week or two before renaming gmp-ecm to ecm so that my replaces=('ecm') has time to take effect? Once that delay has passed I'll remove said line and you'll be able to safely rename gmp-ecm to ecm.

Thanks a lot Maxime!

Just another comment: Please do not rename gmp-ecm into ecm; gmp-ecm is the proper upstream name for that package. The conflict was only between binary names, no need to rename packages too.