[
https://issues.apache.org/jira/browse/MRESOLVER-290?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17633339#comment-17633339
]
ASF GitHub Bot commented on MRESOLVER-290:
------------------------------------------
cstamas commented on PR #218:
URL: https://github.com/apache/maven-resolver/pull/218#issuecomment-1312746749
There's much more to do, but am stopping here due
https://issues.apache.org/jira/browse/MRESOLVER-295.
My plan is following:
* component should use component (even if trivial as Slawek noted) to
perform IO. Reason: components can be swapped out
* split FileProcessor into PathProcessor (as name says, for files) and
ChecksumProcessor (to read write checksums ONLY, to not mix concerns). This
opens up the possibility to implement different ChecksumWriters
(BSD/GNU/whatever), as currently we read BSD/GNU/resolver-native but write only
resolver-native format.
* but, due module layout (SPI not depending on Utils, and both Impl and
Connector depending on SPI but Connector not on Impl) this needs a bit more
work: see how many implementations of FileProcessor we have across, exactly for
this reason. So, I want to do something with it.
* direct use of static helper reduces testability, there is a nice test
where file copy fails, that cannot be done if directly static helper used.
> Improve file handling resolver wide
> -----------------------------------
>
> Key: MRESOLVER-290
> URL: https://issues.apache.org/jira/browse/MRESOLVER-290
> Project: Maven Resolver
> Issue Type: Task
> Components: Resolver
> Reporter: Tamas Cservenak
> Assignee: Tamas Cservenak
> Priority: Major
> Fix For: 1.9.1
>
>
> Use new atomic primitives (utils) to make sure files are fully written. Also,
> tighten up IO, as some parts are "forgiving" re IOexceptions (like writing
> checksum files to disk?), but this relaxedness is wrong, as if you have IOex
> while writing to your disk (local repo) is usually a symptom of some real
> problem, from permissions mixup to possibly faulty hardware.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
