Bug - Not A Bug Tables in the gCLI mutliplying

It is a duplicate; the other customer got an infinite loop and tried a workaround that resulted in multiple tables being displayed.

I am not happy that the comments say things like "this will be fixed in Version 9, since Version 7 is in Sustaining". Does "Sustaining" continue to release updated builds?

Seriously? We're waiting for version 1.9 to fix this? When even version 1.8 has not been officially released?!?! I am not pleased with the company.

Posted for reference: JDK-8048110

Veracity said:

I am not happy that the comments say things like "this will be fixed in Version 9, since Version 7 is in Sustaining". Does "Sustaining" continue to release updated builds?

Click to expand...

Arguably yes. It seems that 1.7 is the official release which only has updates for reasons of stability, security or serious bugs. That does fall into the domain of "sustaining."

I suspect that version 1.8 is the beta test version which can have new features and changes which may produce instability. Unfortunately this fix has been pushed all the way down to version 1.9, the alpha release, which is for testing of new features and other alterations which are not reliable enough to expose to the public.

That surprises me, since it was a regression: a released bug fix broke previously working code. If it were I making the call, I'd put it back into version 7, to restore functionality which previously worked in that release.

I completely agree with your call. That's why I'm saying that I'm not pleased with them.

Because it's always such maintenace delight to fix something in a branch that's two major releases behind your current "in development" code version, and then forward-port it to at least two branches.

It's more common to fix things in the last couple of major production releases, though. Or does Java use beta in the same sense as KoL does!

xKiv said:

Because it's always such maintenace delight to fix something in a branch that's two major releases behind your current "in development" code version, and then forward-port it to at least two branches.

Click to expand...

Sarcasm does not become you.

They broke something midway through the maintenance period of version 7; it worked in 7u55 and was broken in 7u60. The current released version is 7u67. The initial bug report said that it was a deal killer for the customer; they cannot proceed with their development in the face of this bug. There is no workaround for the bug other than to not upgrade past 7u55. Now, an additional customer (us) reported also being bitten by the bug and finding no workaround.

And yet, for some reason, they classify the severity of the bug as "3".

Given all that, they target it to be fixed in a version which will be released in, oh, 2 years?

I have no sympathy for the "maintenance delight" argument.

I've kind of lost track of everything but since the Java bug seems unlikely to be addressed any time soon, is there a code change or workaround that we could implement to fix this for users who are on something newer than 7u55? If not, for either technical of philosophical reasons, should we consider an "official" position that mafia requires 7u55 or older and write something in bold red letters to the gCLI if the user is not meeting that requirement?

As we have said repeatedly on this thread - and as has been pointed out to Oracle in two separate bug reports - there is no "workaround". Unless by "workaround" you mean "don't use HTML tables". That's a non-starter.

We also don't "require" that 7u55 or earlier. If you don't, there will be a cosmetic bug for the handful of commands (and user scripts) that generate HTML tables in the gCLI.

Fortunately the cosmetic bug does not make tables unreadable. It's more like, "WTF? Why is all this garbage after the usable information?" It will "annoy" users, but there's nothing that we can do about it so if anyone ever asks why this problem exists we can just copy/paste the link, JDK-8048110, and ask them to bug javaJDK for a fix in the current version of java.

It's worth noting that they haven't even fixed in in java 9 yet. It's still unresolved. :sigh: Maybe once they fix it in java 9 they will consider the virtues of porting it up two branches.

Bale said:

Fortunately the cosmetic bug does not make tables unreadable. It's more like, "WTF? Why is all this garbage after the usable information?" It will "annoy" users, but there's nothing that we can do about it so if anyone ever asks why this problem exists we can just copy/paste the link, JDK-8048110, and ask them to bug javaJDK for a fix in the current version of java.

It's worth noting that they haven't even fixed in in java 9 yet. It's still unresolved. :sigh: Maybe once they fix it in java 9 they will consider the virtues of porting it up two branches.

Click to expand...

Java support has really gone downhill since Oracle bought Sun. It's really a shame but there is literally nothing that can be done by the KoLmafia team.

So, looks like this will be fixed with Java 7u80, now. However, from what I've been able to tell, that's not planned for release any time soon. The reason I'm bumping, though, is that I seem to have stumbled upon a workaround. With some of the recent changes to font size display in mafia, I ended up on the preferences page while I had the CLI tab open. I saw that the CLI changed font immediately on clicking, and as luck would have it, I had a "broken" table visible. Much to my surprise, the table got "fixed" when I changed font size.

In the attached patch, I've taken the line I found triggered the update, and inserted it into where I think all table output would pass through. I'm not familiar enough with Java or the mafia code to say if this is the right way to do it, or if there are consequences I hadn't thought of. But with that patch my CLI tables now look normal and I haven't noticed anything behaving oddly.

Edit: OK, there is some oddity, in that it seems to get very resource hungry when there's a lot of text scrolling by at once. I updated the patch with a test for table output, since that's the only time the output goes bad.

That's fantastic news! I'm happy they finally fixed it even though it isn't getting to release soon. I did find this on the expected date:

Just wanted to confirm that 1.8.40 DOES fix this Java bug, so now we can tell people to update their Java, not just their mafia. Yay...

Very good. Finally!