search

cschwan / sage-on-gentoo

(Unofficial) Gentoo Overlay for Sage- and Sage-related ebuilds
84 stars 26 forks source link

update requires dev-lisp/ecls-16.1.3-r100? #534

Closed strogdon closed 5 years ago

strogdon commented 5 years ago

Updating a Gentoo box with s-o-g 8.6 installed. The update requires the subject ecls

These are the packages that would be merged, in order:

Calculating dependencies... done!
[ebuild  r  U ~] dev-lisp/ecls-16.1.3-r100 [16.1.2-r1]
[ebuild     U  ] dev-libs/expat-2.2.6 [2.2.5]
[ebuild     U  ] net-misc/curl-7.64.0 
...

Is this safe? Is it possible that >=dev-lisp/ecls-16.0.0::sage-on-gentoo in package.keywords/sage-8.6 is overriding ~dev-lisp/ecls-16.1.2 in the ebuild?

strogdon commented 5 years ago

sage was installed with emerge --oneshot so perhaps that's why the newer ecl is being pulled in.

kiwifb commented 5 years ago

Not safe yet. sage does hard depend on 16.1.2 - We may be able to move with sage 8.7 the tiket should have positive review.

strogdon commented 5 years ago

So is it because I did a emerge --oneshot in installing sage that now the subject ecls is being pulled in. Or do I just need to mask it (which is what I have done).

kiwifb commented 5 years ago

Honestly I am not sure. oneshot doesn't prevent portage to look at dependencies of the stuff in your list of software. It may have fallen in a hole in the dependency graph. What did trigger the ecls rebuild?

strogdon commented 5 years ago

I'm not sure what triggered it.

equery d ecls
 * These packages depend on ecls:
sci-mathematics/maxima-5.41.0-r3 (ecls ? dev-lisp/ecls)
sci-mathematics/sage-8.6 (~dev-lisp/ecls-16.1.2)
virtual/commonlisp-1 (dev-lisp/ecls)

so I don't think it should have been pulled in. I stopped the build when I noticed the newer ecls was being installed. ecls was in my world file. I removed it but that didn't affect which ecls was built. It's now masked so I'm good.

strogdon commented 5 years ago

closing. this must be something on my end?

kiwifb commented 5 years ago

Not sure really. The "r" tag can mean that something is causing a rebuild but at the start like here it is more likely that it the cause of re-builds. Usually the rebuild is to do with subslots. So a sage rebuild may not have been triggered because it is the same subslot. If it had been triggered then sage reverse dependency would have been taken into account. I am wondering about the virtual/commonlisp and whether that was the cause.

strogdon commented 5 years ago

This fixed things here

#>=dev-lisp/ecls-16.0.0::sage-on-gentoo
~dev-lisp/ecls-16.1.2::sage-on-gentoo

in sage-on-gentoo/package.keywords/sage-8.6. But this change is not needed in my Prefix (sage-9999 installed). Strange? I removed the virtual/commonlisp - no change.

kiwifb commented 5 years ago

That wouldn't help people on ~arch and they haven't complained with that particular problem (at least it hasn't reached me). And of course your prefix is basically ~arch. I'll have a closer later when I am properly back from my conference.