[
https://issues.apache.org/jira/browse/MWRAPPER-123?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17813526#comment-17813526
]
ASF GitHub Bot commented on MWRAPPER-123:
-----------------------------------------
bmarwell commented on PR #117:
URL: https://github.com/apache/maven-wrapper/pull/117#issuecomment-1923203779
> My test:
>
> ```
> hash_string() {
> str="${1:-}" h=0
> while [ -n "$str" ]; do
> c="${str%${str#?}}"
> h=$(( ( h * 31 + $(LC_CTYPE=C printf %d "'$c") ) % 4294967296 ))
> str="${str#?}"
> done
> printf %x\\n $h
> }
> ```
>
> @bmarwell can you try on AIX?
sadly...
```
# vi test.sh
# ksh test.sh
"https://reopsitory/repository/maven-central/org/apache/maven/apache-maven/3.9.6/apache-maven-3.9.6-bin.zip"
printf: 2589392689: A return value of a math subroutine is not within
machine precision.
7fffffff
```
> only-mvnw fails on AIX' ksh implementation
> ------------------------------------------
>
> Key: MWRAPPER-123
> URL: https://issues.apache.org/jira/browse/MWRAPPER-123
> Project: Maven Wrapper
> Issue Type: Bug
> Affects Versions: 3.1.1
> Reporter: Benjamin Marwell
> Priority: Urgent
>
> the script {{only-mvnw}} contains this snippet in a line to calculate a hash:
> {code:bash}
> printf %d "'$str"
> {code}
> The idea is to convert {{$str}} to a number, where {{'}} has a special
> meaning for {{%d}} conversion. However, there are two problems with it:
> 1. It does not work on ksh on AIX (because this conversion is not posix
> compliant), so the script will not run on AIX.
> 2. It will only use the first letter, which is alway {{h}} (from the string
> always being a valid URI).
> For now, replace it with the ASCII constant 104 (ASCII value of h). The
> intended behaviour can be restored in another PR if feasible.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)