On Wed, 15 Apr 2026 15:53:29 GMT, Alan Bateman <[email protected]> wrote:

>> The most correct technical name for this would be "trust anchors", so I can 
>> consider changing the option name to --trust-anchors and changing wording to 
>> "the keystore containing trust anchors". However, keep in mind that 
>> developers have gotten used to the `cacerts` keystore name, so there may be 
>> some confusion. You do need to know what is already in the `cacerts` 
>> keystore to effectively use this option, at least with a JDK based on the 
>> OpenJDK source.
>> 
>> As for `keytool`, I don't think we can or should change the `-cacerts` 
>> option name. My guess is you are concerned with the words "the `cacerts` 
>> keystore" in various places and the pathnames under the `-cacerts` option.  
>> It is kind of hard to word around this and make the docs easy to understand 
>> though. The `cacerts` keystore is an essential important part of developing 
>> and debugging secure Java applications. But I can file an issue to see if we 
>> can abstract this a bit.
>
> Just to be clear, I'm not not suggesting changing the keytool -cacerts 
> option, my comment was about the keytool man page. I initially thought the 
> man page referenced lib/security/cacerts but reading it again, it's the 
> security properties file. We may have forgotten to change the file paths to 
> conf/security.

Just to conclude on this. I think the --cacerts option and have it take a list 
of aliases is fine. 

For the wording then it might be a bit clearer to say "with only the 
certificates ..." rather than putting "only" at the end.

Part of also thinks that "<alias> is the name of an alias in the cacerts 
keystore" should say the cacerts keystore in the java.base module (as this is 
where jlink gets the certificates).

Summary - useful feature when creating a run-time image for a specific 
application or usage.

-------------

PR Review Comment: https://git.openjdk.org/jdk/pull/29700#discussion_r3094531512

Reply via email to