Introducing the Rear Window blog

@trixieQuote:
The link to your blog just goes to a blank page. Tried the link on to the boot on your main Ko-fi page and also get a blank page. Just an fyi.

Nevermind. It's working now.

@trixie

Quote:
Unless I'm doing something wrong, there seems to be a bug in the new version of Rave. If I bring up the Settings requester and then click on either "File Requester" or "Advanced" the program stops responding, along with much of the OS. I have to reboot to recover.

@trixie

Thanks for the update and blogpost!

@msteed

Quote:
Hmm, sorry about that, but this is strange because I haven't touched this part of the program since the previous version at all.

I test Rave on three different systems: X5000/020, Sam440ep-Flex, and WinUAE running AmigaOS4 FE for Classic. All of these machines are at Update 3, and none of them shows the problem you've described. So I don't know what to say

@trixie

We may be getting a bit off-topic here, but this thread has become kind of a place to discuss Rave in particular, as well as the blog in general. So...

Quote:
I was afraid of that- you do quality work, and it was hard to imagine you missing something like this. Nonetheless, it's entirely repeatable on my system (which is an X1000 running Update 3 and Enhancer 2.2).

I spent some time investigating the problem today, and it's rather bizarre.

First, I booted up my Update 2 partition -- which I made for just such a purpose prior to installing Update 3 -- and confirmed that Rave 1.9 works fine under Update 2- clicking on "File Requester" or "Advanced" just brings up the expected page of controls.

Back under Update 3, I unpacked the Rave 1.9 archive into RAM:, and discovered that Rave 1.9 works properly when run from the RAM disk, even under Update 3.

Thinking that perhaps my hard disk installation got corrupted somehow, I then drug the Rave drawer from RAM: to the hard drive (a different partition than my normal installation), and when run from the hard drive Rave once again hung when I tried to display the same two settings pages.

I next (after rebooting) installed Rave 1.8 to the same partition on the hard drive, and confirmed that it worked properly, even when run from the hard disk.

So, to summarize:

- Rave 1.8 works properly under Update 2 and Update 3, from RAM: or from the hard drive.

- Rave 1.9 works properly under Update 2 from the hard drive (I didn't try RAM:), and works properly under Update 3 when run from RAM:.

- Rave 1.9 does not work properly under Update 3 when run from the hard drive. (I use SFS2 on my hard drive partitions, not that the file system should make any difference.)

Some experimentation with Sashimi revealed that there is in fact a GR when Rave crashes, but due to side effects of the crash the GR window never appears. I was able to capture the crash log to disk:


It looks like the crash occurs while setting up the gadgets for the selected page, which makes sense. No idea why it occurs when run from the hard disk and not when run from RAM:.

At this point I'll toss the ball back to you, and see if the crash log provides any clues.

@msteed

Let me contact you off-forum and send you a debug build for more information.

@trixie

I see the same behaviour on my X1000 as msteed, although this was (so far) just a quick test on my beta install.

If you need it, I can make a serial log, and also try it on a public install and/or on my X5000/40 if relevant. You know my contact details .

Best regards,

Niels

On my side, X1000 also, I've got a complete freeze of the app if I play with the preferences, Once I click on the File Requester option.

Kind regards,
Xavier.

@trixie

Thanks for the update. Working fine here on my MicroA1.

---
redfox

Quote:

+1. I hadn't realized there was still a need for Sashimi (or even an OS4 port) in the era of Grim Reapers and serial output.

@trixie

Quote:
Standing by...

@tao

Quote:
The crash prevented the Grim Reaper window from opening, and I don't have another machine handy to capture serial debug output. So I use Sashimi to redirect the serial output to a console window, or in this case a disk file. That won't work if the system goes down completely, but that wasn't the case here.

@trixie

Same happens here with X1000.

When launched from RAM I can play in settings, when copied to SSD, opening Rave's settings window locks the Tool / App.

The system continues to work in background, I can move the mouse but can not quit Rave.
No GrimReaper. OS4.1FE update 3

On XE the issue is not present. Nor on Sam440.

I am happy you are in the OS4 Dev Team!

@trixie

The latest version of AmigaDiskBench (v. 2.5.3) seems to have a very similar bug as Rave: When I run the program, select the Disk Info tab, and then click on "Fixed Drives/sb600sata.device" in the listview the program stops responding, along with much of the OS. (There is a recoverable DSI when I first run the program; that DSI doesn't seem related to the lockup bug.)

Like Rave, the crash happens when the program is working with ReAction gadgets, and the stack trace shows the crash occurs during a series of nested calls to the layout.gadget, followed by a call to the button.gadget. The first part of the crash log, including the general info and the register dump, seems to go missing. But what remains is enough to get an idea what was happening when the crash occurred:


The exact code where the crash occurs is different than Rave. There are some other differences as well: unlike Rave, the crash occurs under FE Update 2, as well as under Update 3. And unlike Rave, the crash occurs even if AmigaDiskBench is run from the RAM disk. While the crash occurs most of the time, it occasionally doesn't, unlike Rave where it is always repeatable. Since I have only an X1000, I can't say if the crash only occurs on the X1000, as it does with Rave.

AmigaDiskBench isn't your responsibility, but the fact that a very similar bug occurs there is some evidence that it's not caused by your code doing anything wrong, and may in fact be a system problem of some sort. And perhaps the differences between the Rave crash and the AmigaDiskBench crash provide some clues as to where the problem lies.

@all

With the help of Niels Bache, the X1000-only crash has been fixed for Rave 1.10.

@trixieQuote:
Thank to both of you!

@trixie

Quote:
With all the weird symptoms, I'm curious what the problem turned out to be.

@msteed

Quote:
It was a race condition. More details in my next blog post