Db101 betatesters needed

I tried to debug a binary compiled by ECX (AmigaE compiler) with debug information enabled. db101 displayed no source code lines. Does it mean a debugger contains code to interpret C code only?
I had the impression, using some kind of offset could make it language agnostic.

@Elwood

ECX inserts stabs based debug info into the executable?

If it doesn't all bets are off...

@broadblues

Damn, I missed the part in bold in the manual
STEP=STEPDEBUG/S - Makes it possible to debug code with a debugger.
Not implemented yet.

@Raziel

I assume you mean, that you clicked on the 'params' entry in the 'Variable' list column.

I have built a fresh version with a minor change. I hope you can test it a little bit.

https://www.justbeamit.com/87zvy

@alfkil

Quote:

@samo79

Hmm... Is it impossible to find a file transfer service that works on OS4.1?

It didn't work for me either, and I'm running Chrome under Windows 7, so it's not an Amiga thing.

Simon

@alfkil

Yep, link is dead.

I use TinyUpload if i need to.
It's perfect for such "small" files and it's free too (50MB)

@Raziel

Here is the TinyUpload link: http://s000.tinyupload.com/index.php?file_id=37399584747346489462

Hope it works.

@alfkil

Download worked and the new version let me open the sub item, thanks a lot for the fix

Is there a chance i'll be able to save the stacktrace to a file in a future version (for record and to ask other devs)

@alfkil

If i "Kill" a crashed program i get a GrimReaper



Yep, using the debug kernel here...

@Raziel

'Kill' is really an unsupported feature. There is not way to do it in a system compliant way. So don't expect it to work.

All I'm getting from that TinyUrl link is a load of shit popups asking me to install Chrome plugins.

EDIT:

Well, there's a first. I tried with Odyssey and the link worked. I think that must be the first time ever an Amiga browser succeeded where a mainstream browser has failed.

I'll test it tommorrow.

Simon

@Rigo

Quote:Really? It fails every time I try; I just tried again, and no luck. It tells me the file already exists (it doesn't), and I choose Resume. Then it says it failed.

(The page also says you can't download it without signing up - no way ...)

Best regards,

Niels

@nbache


Quote:

That is an 'advert'. One reason I hate those sites, is the phishing style adverts on them. I really wonder why the days of dropbox, google drive et al why anyone needs to use them?

@alfkil

Just tried to attach db101 to a runing process (AOrgansier) that I'd maged to get inot aloop.

Two issues :

1. The list of processes is uninformative. Too manay "Background CLI" entries with no other info. Address and prgram running in that CLI where possible woud be more than helpful.

2. Guessed the right one based on order of starting prgrams I got a message AllocMem failed followed by a DSI when loading debug data, ignoring the DIS produces a string of AlloMem failed outputs....

BTW if it took the same sort of argumenst (or a subset of) as GR sends to gdb we could rename it GDB and perhaps start it from a GR?

@nbache
@Rigo

Thats why i use a custom HOSTS file which suppresses all of tose nasty popup/phishing/spam stuff.

The site looks pretty clean when i use it

@broadblues

I haven't tested the 'attach' mode for a long time. It is possible, that something changed in the os, that needs fixing in db101. I'll try and find time to look at it.

It would be great, if we could replace GDB. I am not sure that I am completely up to date featurewise (you mentioned a few things, that could be interesting to look at), but it would probably create a slightly more inviting atmosphere for the average user.

@alfkil

I have done a small update, where I have enabled clicking the entries in the stacktrace to display the parameters and local stack variables from selected entry.

http://s000.tinyupload.com/index.php?file_id=05112804899334529654

I am not sure it is optimal. It works, but there are certain things, that bother me. Like the intuition of what is going on. Does the user know, that what he is seeing, is not the actual current function but the one selected in the stacktrace window? What about entries, that do not show correctly when selected this way - like static entries and pointer content? Would it be better to show these infos in a separate window with a clear warning, that what you is seeing is not necessarily, what used to be there? Or is it ok? Does one know that much, if one is reading stackframes from a crashed app?