Summary: | sys-apps/portage-2.1.4.4 doesn't verify checksums after redownloading a broken file with FEATURES=-strict | ||
---|---|---|---|
Product: | Portage Development | Reporter: | Petteri Räty (RETIRED) <betelgeuse> |
Component: | Core - Ebuild Support | Assignee: | Portage team <dev-portage> |
Status: | RESOLVED FIXED | ||
Severity: | normal | Keywords: | InVCS |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Bug Depends on: | |||
Bug Blocks: | 216231 |
Description
Petteri Räty (RETIRED)
2008-03-26 13:55:26 UTC
So, the expected behavior is that it should show a warning message and continue despite the invalid checksum? (In reply to comment #1) > So, the expected behavior is that it should show a warning message and continue > despite the invalid checksum? > It should fail like it does when the file is not already in DISTDIR. Hmm, are we talking about behavior with FEATURES=strict enabled or disabled? Perhaps the bug summary should say FEATURES=strict instead of FEATURES=-strict? If you are trying to confuse me then you have succeeded. :P from make.conf.5: strict Have portage react strongly to conditions that have the potential to be dangerous (like missing or incorrect digests for ebuilds or distfiles). So, shouldn't it ignore an incorrect digest if "strict" is disabled? Historically "strict" enabled checksum checks for the manifest types EBUILD and AUX (and MISC in some versions), but DIST files were always checked independent of "strict". It doesn't make sense to skip checksum verification of distfiles in this case anyway as it only delays the failure tilll the checksum check before unpack, while for other types the check is skipped completely with "-strict". Okay, it's fixed in svn r9519 so that distfiles are always checked in any case. This is fixed in 2.1.5_rc1. |