On Tue, 14 Sep 2021 13:28:24 GMT, Daniel Fuchs <dfu...@openjdk.org> wrote:
>> There has been no change in how we deal with this aspect. The existing
>> specification (stated in the `load` method) says:
>>
>>
>>> * Properties are processed in terms of lines. There are two
>>> * kinds of line, <i>natural lines</i> and <i>logical lines</i>.
>>> * A natural line is defined as a line of
>>> * characters that is terminated either by a set of line terminator
>>> * characters ({@code \n} or {@code \r} or {@code \r\n})
>>> * or by the end of the stream. A natural line may be either a blank
>>> line,
>>> * a comment line, or hold all or some of a key-element pair. A logical
>>> * line holds all the data of a key-element pair, which may be spread
>>> * out across several adjacent natural lines by escaping
>>> * the line terminator sequence with a backslash character
>>> * {@code }. **Note that a comment line cannot be extended
>>> * in this manner;**
>> (emphasis on that last sentence).
>> I'll anyway go ahead and add new tests around this to be sure that this
>> works as advertised.
>
> Oh - great - thanks - verifying by a test that it also applies to the comment
> specified by `java.util.Properties.storeDate` would be good.
Done. A new test `StoreReproducibilityTest#testBackSlashInStoreDateValue` has
been added in the latest updated version of this PR. This test passes. Plus I
checked the written out properties file, from these tests, for such values in
`java.util.Properties.storeDate` and the content matches what the spec says.
Just for quick reference - a run of that test case with the
"newline-plus-backslash...." system property value (cannot paste that exact
string value from that test case because GitHub editor is messing up the
special characters) generates output like below:
#some user specified comment
#newline-plus-backslash\
#c=d
a=b
-------------
PR: https://git.openjdk.java.net/jdk/pull/5372
