atom-9999::mva fails to emerge at install phase

ExecutorElassus

So, I got here by several steps:

using app-editor/atom-1.32.0_beta3 from the cynede overlay, I was unable to update dependent packages: the file /usr/share/atom/resources/app/apm/node_modules/npm/node_modules/npm-lifecycle/node-gyp-bin/node-gyp refused to execute with a "Permission denied" error. So I thought this was just something wrong with that version (the file, incidentally, also points to the wrong directory to find the linked node-gyp.js executable, but even when this is fixed it fails out). So I tried the mva overlay, which appears to have a "live" ebuild for atom-9999.

Trying to install that, however, it fails at the install phase with the following error:

John R. Graham · Posted: Fri Oct 12, 2018 3:49 pm Post subject:

Moved from Portage & Programming to Unsupported Software. The cynede overlay is not officially supported so the thread fits better here.

- John
_________________
I can confirm that I have received between 0 and 499 National Security Letters.

ExecutorElassus · Posted: Fri Oct 12, 2018 5:56 pm Post subject:

well, technically we're talking about the mva overlay now, since that's the one I can't emerge. But since that one isn't officially supported either, iirc, I guess it doesn't much matter.

Any guess what this error is?

Cheers,

EE

ExecutorElassus · Posted: Tue Oct 16, 2018 1:37 pm Post subject:

I've been digging into this a bit, and now I have a peripheral question: it may be the case that atom-1.32.0 requires a more recent version of nodejs, but that would require using openssl-1.1*, which is currently masked. So here's my next question: how likely will unmasking openssl-1.1.0i (which is what's in the portage tree right now for the -1.1* series) børk my system? The bugreports strongly suggest the likelihood is high, but maybe it's not so terrible?

Cheers,

EE

fedeliallalinea · Posted: Tue Oct 16, 2018 1:47 pm Post subject:

You can check if there are packages installed in your system with a open bug for openssl-1.1

EDIT: openssl-1.1 is hard masked, then the chance of having problems is high
_________________
Questions are guaranteed in life; Answers aren't.

Hu · Moderator Joined: 06 Mar 2007 Posts: 21586

We have somewhat routinely see forum threads from people who had a problem that was traced to them unmasking OpenSSL 1.1. In those cases, they did it unknowingly because they blindly trusted autounmask. You would be in a slightly better position since you would know to be suspicious of OpenSSL-related failures. If you retain a tbz2 of openssl-1.0.x, you should be able to back out if the situation goes badly. There might still be a fair amount of rebuilding to get all packages back to the 1.0.x subslot though.

ExecutorElassus · Posted: Sat Oct 20, 2018 7:45 pm Post subject:

addendum: it turns out that the problem I have with atom only occurs on the 1.32.* branch. The 1.31.* version (which appears to be the last that appears to require USE="python_targets_python2_7") works fine.

So, I suspect that whatever changed for the 1.32.* branch, be it python 3, or something else, is preventing the gyp-rebuild process from working properly.

But since the workaround is just to use atom-1.31.1, I can just do that.

Cheers,

EE