[ 
https://issues.apache.org/jira/browse/MRESOLVER-273?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17609250#comment-17609250
 ] 

ASF GitHub Bot commented on MRESOLVER-273:
------------------------------------------

cstamas commented on PR #194:
URL: https://github.com/apache/maven-resolver/pull/194#issuecomment-1257280023

   > As far as I understand the Provider approach now is that the actual 
mappers might be incomplete and the provider will compose them properly to 
usable mappers?
   
   Exactly, not "incomplete" but they are now "simple" each doing one simple 
thing. And they are NOT components anymore, as providers are actually composing 
them. This is the "cleanup" I refer to in PR description: before, NameMapper 
implementations were highly duplicated, for example I'd need to duplicate file 
basedir resolution for any new file frinedly mapper. But with this PR, not 
anymore, all you need is new provider and create a new "chain" (well, not like 
real servlet chain, but you get the point)




> Create more compact File locking layout/mapper
> ----------------------------------------------
>
>                 Key: MRESOLVER-273
>                 URL: https://issues.apache.org/jira/browse/MRESOLVER-273
>             Project: Maven Resolver
>          Issue Type: Improvement
>          Components: Resolver
>            Reporter: Tamás Cservenák
>            Assignee: Tamás Cservenák
>            Priority: Major
>             Fix For: resolver-next
>
>
> Current FileGAVNameMapper "mimics" loosely local reposiory layout, uses long 
> paths, hence (relatively) big strings etc. 
> More compact layout would be just to hash strings like "a:G:A:V" and 
> "m:G[:A[:V]]" for artifacts and metadata instead, and create 2 level deep 
> hashed storage.
> Problem with "loose" layout is that they do end up as files in OS, while 
> these would be much shorter.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to