Red dot breakpoints do not stick in v9

[dunbarx]

I heard they did, so I downloaded 9.0.3.

They do not. Rats.

And v9 quit a couple of times, so I reverted back to 8.1.10. V.8 does quit now and then, but the fact that v.9 quit within half an hour makes me nervous.

Any comments on either of the above?

Craig

[FourthWorld]

No recipe, no dice.

[bogs]

I'm not sure I understand , Craig is talking about soft break points. What recipe is there for setting a soft break point?

Any comments on either of the above?

Well, on 'nix the soft break points stay in any version I've used, far as I can tell.

The 'IDE quitting out' has prevented me from even testing a distro later than 8.1.2 (on my actual box), beyond that version on my (unsupported) distro, you can do many many things, if it launches, but programming isn't one of them.

[dunbarx]

Richard.

No recipe, no dice.

For the LC crash? For the dots, I simply took a couple of stacks, opened them in v.9 and started playing in the SE. If I move lines, add lines, delete lines, whatever, with a red dot in place, then when I run that handler, the breakpoint is ignored.

Just like it always was.

But in a NEW STACK MADE IN v.9, the red dot seems to stick.

That is progress, but of little value to me right now. And I am skittish about just changing due to stability.

Craig

[jacque]

The crashes after debugging are an ongoing mystery I've mentioned on the mailing list. There is no recipe except that it happens after tracing through a handler in the debugger. I submitted 5 crash logs and Panos was going to see if he could glean anything from them. I've since begun collecting another set.

Other symptoms are the message box suddenly being unable to find things, like a handler in the frontmost card or an object you're looking at. Keyboard shortcuts also stop working. When these symptoms appear you are headed for a crash soon. The behavior can't be reproduced on demand and it's possible to trace and debug for some time successfully and then suddenly run into the problem.

If you aren't debugging, LC 9 is quite stable.

[dunbarx]

Jacque said:

If you aren't debugging, LC 9 is quite stable.

Who's says she doesn't have a sly, slightly evil, understated sense of humor?

Craig

[dunbarx]

I see, in the v.8.2 DP2 release notes, two items that profess the repair of the red dot issue.

Anyone use these versions reliably? I would LOVE to be able to trust the little buggers. Right now I virtually never use them, rather writing breakpoint commands all over the place.

I ask because maybe if I am still in v.8x, my existing stacks will inherit the repair. As I said above, a stack made in v.8 will not do so if opened in v.9.

Another option, as Jacque mentioned, is simply write perfect code, so no debugging is ever necessary. But what is the fun in that?

Craig

[FourthWorld]

I'm not sure I understand , Craig is talking about soft break points. What recipe is there for setting a soft break point?

I'm hoping there's a recipe for reproducing the bug. I've been using soft breakpoints all day without issue in v9.0.3.

[jacque]

Try stepping through and into several chained handlers and aborting without completing some. I've crashed right after an abort, but more often the other symptoms appear some time later. If the message box gets confused or keyboard shortcuts stop working, you're close to a crash.

I spent a couple of days doing layout and writing a few handlers without any problem, but when I started intensive debugging it happened. Sometimes immediately, sometimes after a day. There's no pattern but it looks like tracing through the debugger does it.

It's possible this is a Mac thing only.

[SparkOut]

Red dots being ignored used to be a thing on Windows too, but it hasn't been a problem for a number of versions now. Sometimes the dots go orange and there's a mismatch in line numbers when editing, and it takes a few goes to "apply" for them to get back in sync and work as breakpoints again. I've noticed the line numbering stay the same while scrolling the code up and down and taking another go to apply, even without editing. Maybe it's connected to the line number formatting code not matching the red dot position?
But starting in 9.0.3 I have experienced the IDE having successfully edited, applied and saved scripts. Then switching to browse mode and testing reveals no edits made, and examining the script shows it has reverted completely. There is no recipe, it didn't start immediately, I had already been coding on different days, then it suddenly started happening. Maybe 6 times out of 10. If I edit, save, close LC and open again the edits are reliably correct. If I try editing, saving and testing, it's pretty bad. I mention this in case it is somewhat related (to the line number formatting part of the script editor) or just general IDE .

[Simon Knight]

Hi,

I have experienced the IDE having successfully edited, applied and saved scripts. Then switching to browse mode and testing reveals no edits made, and examining the script shows it has reverted completely.

I've seen this when using 9.0.2 on a mac. It was related to debugging but seems random. I have also noticed that at times the line numbers freeze while the code scrolls, sometimes they update and other times they don't. Whenever I witness this I know it is time to quit and restart.

Lastly, sometimes when building a standalone the IDE goes into a loop of error messages posting messages alone the lines of "a stack of the same name is already in memory", sometimes the purge works sometimes not, when it fails the IDE crashes. In any case there was not a stack of the same name present before the loop, if it exists it is created by the IDE. Again I have not found a way to reliably reproduce the error.

Oh and another thing.... I opened a presentation created by Scott Rossi back in 2014 which is written in Livecode and contains many cards. However, this stack and its cards sends the Project Browser potty :

Otherwise - no problems

best wishes

Simon

[FourthWorld]

I've seen the line numbers out of sync, but thought perhaps that might be just in their Linux engine. Still haven't seen the breakpoints issue, but I rarely use those so I don't have a recipe.

v9.0.4 Stable was released today. Maybe give that a go and if you see these issues again let's see if we can pin down a recipe.

[Simon Knight]

I'm just downloading 9.0.4 and will report any strangeness .

Also, just submitted a bug report re the project browser : #22058
https://quality.livecode.com/show_bug.cgi?id=22058

[dunbarx]

Just tried v.9.0.4. Red dots still do not stick.

A shame, because these could be such a time saver, and they are easily spotted without having to refer to the "breakpoints" tab in the SE. They cannot be used, because it is more work to go back and constantly reset them than it is to type hard breakpoint commands inLine.

I guess this is just not a priority with the team.

Craig

[FourthWorld]

Just tried v.9.0.4. Red dots still do not stick.
...
I guess this is just not a priority with the team.

If memory serves, their limitation has been like mine: I haven't been able to reproduce this.

What is the recipe for this? Once I can reproduce it I can help steward the bug report.