World: r3wp

The fix was is alpha 98.

was in

Something changed between A107 and A108.

A lot changed. What do you mean?

In this specific case.

Ah, I see.

>> make error! [type: 'foo id: 'bar]           
** foo error: (improperly formatted error)

Shouldn't have tested with type: 'script earlier :)

** script error
 and "** Script error" are fatally similar.

If the error checking of make error! itself is more strict in A108, 
cool :)

So the change is that `make error!` now is strict in what error types 
it accepts.

Yes, exactly.

Oh, it must have been to error! that was fixed in a98 in the same 
way. Cool.

Or it was "unfixed" somewhen between A98 and A107 :)

Good.

Should be noted on the changelog, though.

You should add some tests to verify that error code checking. I suggest 
TRY/except as a way to format the tests.

Hmm, A98 had the same behaviour as A107, for both `to error!` and 
`make error!`.

['error = try/except [make error! [type: 'foo id: 'bar]] ['error]]

That kind of test would help.

try/except [make error! [type: 'foo id: 'bar]] [true]

Good, good!

A correctly made error is also TRUE? until it is triggered.

Indeed.

So if the tests succeed on TRUE? conditions, rather than #[true] 
ones, your test will fail with false success. Does your test processor 
check for #[true] results, or does it just standard REBOL conditional 
code?

does it just use

It checks for logic?.

IIRC, but I haven't yet read the code in detail. Given that it was 
written by Ladislav, I suspect it is rather advanced.

Yes, it does indeed logic? test results.

But making the tests a bit more explicit does no harm :)

The changes to error creation came with the fix to bug#1593. I put 
a comment to that effect in the ticket.

While you are at it, please reopen bug#1667 and bug#1679, both of 
which are not really fixed in A108.

I am going through the bugs in numeric order, testing them I have 
reopened a few so far.

You can also mark bug#1645 as tested, works for me.

Unless you are an admin, you shouldn't have been able to change the 
status of CC #1647 even if you submitted the ticket, since it was 
marked built. It looks like that CC bug is back.

I can change all tickets I submitted.

Anything else would be stupid.

CC tickets are supposed to only be changeable by reviewers after 
they have been reviewed. Otherwise the reviewing process is broken.

Could you test #1651?

Sorry, I mean #1652.

Not without a hoskit.

Added another comment to #328.

That would slow down decompression quite a bit. It looks like a /zlib 
option would be a good idea, to tell it when to try that method.

Can you determine the length of the uncompressed data using that 
method?

Of course.

But as I mention, if you do that, it's already 90% of what the decompressor 
does.

I'll mention in a comment to that ticket my preferred method of your 
three choices.

Fine, I'll state my preference afterwards :)

But I think the sane choice is pretty obvious :)

Done. Go ahead.