Summary: | sys-apps/portage: triggering a sandbox violation during portage dep generation fails badly: * ISE:write_logfile: unable to append logfile: ... | ||
---|---|---|---|
Product: | Portage Development | Reporter: | Martin Mokrejš <mmokrejs> |
Component: | Sandbox | Assignee: | Portage team <dev-portage> |
Status: | CONFIRMED --- | ||
Severity: | normal | CC: | sandbox |
Priority: | Normal | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Martin Mokrejš
2015-03-19 10:18:03 UTC
i think this is portage not setting up the sandbox env vars properly, and not checking the log status after the depend phase SANDBOX_LOG is set to a relative path ("sci-biology_-_transdecoder-2.0.1"), and the working directory is somewhere weird/bad (/usr/lib64/python2.7/site-packages/). portage should really be setting the sandbox.log to a constant location like $T/sandbox.log even during depend phases. Yeah, the EbuildMetadataPhase class need to be fixed to handle SANDBOX_LOG. I guess we can have it just create a temporary file, since creating directories seems like overkill. The invalid SANDBOX_LOG setting comes from here: pym/portage/package/ebuild/doebuild.py: mysettings["SANDBOX_LOG"] = mycpv.replace("/", "_-_") |