8361972: Clarify the condition of System.console() about standard input/output

[naotoj]

This accompanies the fix for JDK-8361613, which restricts System.console() to return a Console instance only when both standard input and output are connected to a terminal. The change here is solely a specification clarification and tightening of the java.io.Console javadoc to reflect this behavior. We are separating the spec clarification because the fix itself may be backported to prior LTS releases without requiring a Maintenance Review process.

---

Progress

• Change must be properly reviewed (1 review required, with at least 1 Reviewer)
• Change must not contain extraneous whitespace
• Commit message must refer to an issue
• Change requires CSR request JDK-8361981 to be approved

Issues

• JDK-8361972: Clarify the condition of System.console() about standard input/output (Bug - P4)
• JDK-8361981: Clarify the condition of System.console() about standard input/output (CSR)

Reviewers

• Justin Lu (@justin-curtis-lu - Committer) 🔄 Re-review required (review applies to e2a64c95)
• Joe Wang (@JoeWang-Java - Reviewer) 🔄 Re-review required (review applies to e2a64c95)

Reviewing

Using git

Checkout this PR locally:
$ git fetch https://git.openjdk.org/jdk.git pull/26299/head:pull/26299
$ git checkout pull/26299

Update a local copy of the PR:
$ git checkout pull/26299
$ git pull https://git.openjdk.org/jdk.git pull/26299/head

Using Skara CLI tools

Checkout this PR locally:
$ git pr checkout 26299

View PR using the GUI difftool:
$ git pr show -t 26299

Using diff file

Download this PR as a diff file:
https://git.openjdk.org/jdk/pull/26299.diff

Using Webrev

Link to Webrev Comment

[bridgekeeper]

👋 Welcome back naoto! A progress list of the required criteria for merging this PR into master will be added to the body of your pull request. There are additional pull request commands available for use with this pull request.

[openjdk]

❗ This change is not yet ready to be integrated.
See the Progress checklist in the description for automated requirements.

[openjdk]

@naotoj The following label will be automatically applied to this pull request:

• core-libs

When this pull request is ready to be reviewed, an "RFR" email will be sent to the corresponding mailing list. If you would like to change these labels, use the /label pull request command.

[mlbridge]

Webrevs

• 02: Full - Incremental (a3b39049)
• 01: Full - Incremental (e2a64c95)
• 00: Full (4db1950e)

[justin-curtis-lu]

If we are claiming that System.console() should only return a "Console instance only when both standard input and output are connected to a terminal." then is the may wording not restrictive enough? Or is that nuance negligible?

[naotoj]

Intention here is that, we do want to restirct to not return JLine based Console when either/both of stdin/out is redirected, I think we would not want to restrict them as the spec. For example, System.console() in jshell returns non-null, even if the remote JVM is not-attached to a terminal. I will describe this sublety in the CSR

[justin-curtis-lu]

OK, I see your point. It is a hint as to what the behavior of this implementation does, but not enforced in the specification by any means as indicated by the may wording. Nit: if that is the case, then shouldn't there technically be minimal compatibility risk in the CSR, since nothing is being "enforced" specification wise.

[justin-curtis-lu]

I think it might read easier if we reword the structure of the sentence to make it more concise.

E.g.

The virtual machine may not have a console if started automatically, (e.g., by a background scheduler) or if either the standard input or output has been redirected.

[justin-curtis-lu]

New wording lgtm.

[AlanBateman]

Can you clarify what the intent of "it may not have a console" is? Initially I though the intention was that there is no Console when either stdout or stdin is redirected. However, the sentence suggests there may or may not be a Console.

[naotoj]

I was considering the case for jshell, where its System.console() returns a Console instance regardless of the isTTY state:

jdk/src/jdk.jshell/share/classes/jdk/jshell/execution/impl/ConsoleImpl.java

Line 70 in bd55d7a

return console = new RemoteConsole(remoteOutput, remoteInput);

That led me to use may in the updated spec. I'd appreciate your suggestions on wording.

[justin-curtis-lu]

Since the decision of whether a console exists is dependent on the implementation as specified,

Whether a virtual machine has a console is dependent upon the underlying platform

And the existing wording in this paragraph uses open-ended words such as "may" and "typically", it makes sense to me that this change would follow the same style of wording. If you wanted to concretely define what this implementation does, using one of the @impl tags to describe the behavior could be an option?

[AlanBateman]

Okay, so the intent is to specify that it is implementation or configuration specific as to whether there is a Console when standard input or output, but not both, are redirected, is that right? What should users expect isTerminal to return in this case?

[naotoj]

For the jshell case, I think it can return true from its isTerminal() as it is infact interactive environment. That means all the Console implementations return true so that we can replace that "may not have a Console" with "will not have a Console", which is clearer.