SimpleUnitFormat won't parse what it just formatted

[aaime]

The following little program shows a case in which the SimpleUnitFormat does not seem to be able to parse back what it just created:

import javax.measure.Unit;

import tec.uom.se.format.SimpleUnitFormat;

public class RoundtripDemo {

    public static void main(String[] args) {
        String unit = "µmol*m^-2*446.2";
        SimpleUnitFormat format = SimpleUnitFormat.getInstance();
        Unit<?> parsed = format.parse(unit);
        String formatted = format.format(parsed);
        System.out.println("Formatted version: " + formatted);
        Unit<?> parsed2 = format.parse(formatted);
    }
}

The output being:

Formatted version: (1/m²)*446.2·µmol
Exception in thread "main" javax.measure.format.ParserException: Parse Error
	at tec.uom.se.format.SimpleUnitFormat$DefaultFormat.parseProductUnit(SimpleUnitFormat.java:520)
	at tec.uom.se.format.SimpleUnitFormat$DefaultFormat.parseProductUnit(SimpleUnitFormat.java:447)
	at tec.uom.se.format.SimpleUnitFormat.parseObject(SimpleUnitFormat.java:269)
	at tec.uom.se.format.SimpleUnitFormat$DefaultFormat.parse(SimpleUnitFormat.java:787)
	at tec.uom.se.format.SimpleUnitFormat$DefaultFormat.parse(SimpleUnitFormat.java:782)
	at tec.uom.se.format.SimpleUnitFormat$DefaultFormat.parse(SimpleUnitFormat.java:773)
	at tec.uom.se.format.SimpleUnitFormat$FinalDefaultFormat.parse(SimpleUnitFormat.java:803)
	at tec.uom.se.format.SimpleUnitFormat$FinalDefaultFormat.parse(SimpleUnitFormat.java:795)
	at RoundTripper.main(RoundTripper.java:13)

Considering that it might be classpath related, here is what I have:

[INFO] |  |  |  |     \- systems.uom:systems-common-java8:jar:0.7.2:compile
[INFO] |  |  |  |        +- tec.uom:uom-se:jar:1.0.8:compile
[INFO] |  |  |  |        |  +- javax.measure:unit-api:jar:1.0:compile
[INFO] |  |  |  |        |  \- tec.uom.lib:uom-lib-common:jar:1.0.2:compile
[INFO] |  |  |  |        +- si.uom:si-quantity:jar:0.7.1:compile
[INFO] |  |  |  |        \- si.uom:si-units-java8:jar:0.7.1:compile

[aaime]

By the way, I've tried upgrading the dependencies to:

[INFO] |  |  |  |     \- systems.uom:systems-common-java8:jar:0.9:compile
[INFO] |  |  |  |        +- tec.uom:uom-se:jar:1.0.9:compile
[INFO] |  |  |  |        |  +- javax.measure:unit-api:jar:1.0:compile
[INFO] |  |  |  |        |  \- tec.uom.lib:uom-lib-common:jar:1.0.2:compile
[INFO] |  |  |  |        +- si.uom:si-quantity:jar:0.9:compile
[INFO] |  |  |  |        \- si.uom:si-units-java8:jar:0.9:compile

Result is still the same.

[keilw]

You should not use something like "µ" with SimpleUnitFormat.Flavor.ASCII. Have to check the rest in more detail but these two are not compatible. Did you try DEFAULT?

[aaime]

Ah sorry, I actually tried both ASCII and DEFAULT, both fail (the output is slightly different, the substance does not change, the class cannot parse what it just formatted). Let me update the code sample.

[keilw]

It's a little weird, but currently our focus is on Indriya where a sort of round-trip from SimpleUnitFormat to EBNFUnitFormat works: https://github.com/unitsofmeasurement/uom-demos/blob/master/console/basic/src/main/java/tech/uom/demo/format/RoundTripDemo.java (I found the term "Tripper" a little irritating;-)

The same does not work in uom-se, I cannot say, if and who could backport everything that's been improved with Indriya into uom-se, because it will ultimately replace it?
However, if EBNFUnitFormat is applied all the way throughout the demo, it swallows something because Formatted version: µmol·/m². Parsing a "*/" operation leads to an obvious and correct parsing error.

@aaime Would you be able to use EDR level Indriya? Currently si-units-0.9 as opposed to si-units-java8-0.9 is based on Indriya 1.0, not 2.0-EDR. We may skip 1.2 in the upcoming 1.0 version of the RI-based SI Units, if that is desired. Although 2.0-EDR is a prerelease, it is more mature and advanced.

[aaime]

D'oh, sorry, that's not what I mean, had to search the term on urbandictionary.com (not a native english speaker). Updated the example code!
I found https://github.com/unitsofmeasurement/indriya... is that going to be a replacement for "systems.uom:systems-common-java8"?

[keilw]

Not for "systems.uom:systems-common-java8", that (until it may eventually be archived) is supposed to use uom-se, but the modules without the "java8" qualifier, e.g. "systems.uom:systems-common" are backed by Indriya 1.0 already. Indriya is the new Reference Implementation (RI) and when JSR 385 goes live, Indiya 2.0 will be Final. The most recent version is the 2.0-EDR Milestone.

[aaime]

Do you have a guess about when Indiya will be released? Even just a vague guess, weeks, months, years?

[keilw]

It is a pretty clear guess thanks to JCP.next and the SI system undergoing many changes early next year. Even if the SI changes were postponed (unlikely but not impossible) we plan to release 2.0 in Q1 of 2019.

[aaime]

I see you closing the ticket, does that mean we should migrated to Indriya?
I see a 2.0 released a few months ago: https://github.com/unitsofmeasurement/indriya/releases

[keilw]

Yes, if you can and are not bound by a legacy application that uses 1.0 and can't be changed, please do.