Tcl - TIP #86 (was Re: Debug API in the core?)

Author

TIP #86 (was Re: Debug API in the core?)

Eckhard Lehmann

2006-09-27, 7:04 pm

Donal K. Fellows wrote:
> suchenwi wrote:
>
> Note that any script-level hooks might need an extension anyway. It's
> not such a bad thing to need to say [package require debug] though. I
> have seen discussions of such a thing several times in the past too.

I just browsed the TIP list and saw that an appropriate TIP (#86) has
already been made and a patch (against Tcl8.4.9) is supplied. The
status is "draft" - what exactly does that mean... is it what it sounds
like?
Can this one go in 8.5?

BTW, what's the status of TIP #131 :-)?

Eckhard

Donal K. Fellows

2006-09-28, 4:02 am

Eckhard Lehmann wrote:
> I just browsed the TIP list and saw that an appropriate TIP (#86) has
> already been made and a patch (against Tcl8.4.9) is supplied. The
> status is "draft" - what exactly does that mean... is it what it sounds
> like? Can this one go in 8.5?

You think it's in a fit state? Might be time to do a formal review then,
see if someone in the TCT is happy to sponsor a vote.

> BTW, what's the status of TIP #131 :-)?

Doing nothing since my mind is a blank. :-)

Donal.

Eckhard Lehmann

2006-09-28, 8:02 am

Donal K. Fellows wrote:
> Eckhard Lehmann wrote:
>
> You think it's in a fit state? Might be time to do a formal review then,
> see if someone in the TCT is happy to sponsor a vote.

I spent yesterday's evening to patch 8.4.13 and test it. The concept
seems nice and powerful, but the new commands are a bit buggy
(inacceptable crashes in [trace breakpopint] and [trace execution] when
called with wrong arguments and the like).

At the moment, it's not very clear to me how to implement a debugger on
top of it, but that's probably, because I didn't dive deep enough.
My conclusion is (from a users point of view), that it is a good idea
and worth to follow, but it's not finished completely. There has to go
some work into the patch to fix bugs and update it for 8.5.

Eckhard

Jeff Hobbs

2006-09-28, 10:01 pm

Eckhard Lehmann wrote:
> Donal K. Fellows wrote:
>
> I spent yesterday's evening to patch 8.4.13 and test it. The concept
> seems nice and powerful, but the new commands are a bit buggy
> (inacceptable crashes in [trace breakpopint] and [trace execution] when
> called with wrong arguments and the like).
>
> At the moment, it's not very clear to me how to implement a debugger on
> top of it, but that's probably, because I didn't dive deep enough.
> My conclusion is (from a users point of view), that it is a good idea
> and worth to follow, but it's not finished completely. There has to go
> some work into the patch to fix bugs and update it for 8.5.

TIP #86 had some flaws. We (ActiveState) are working on one that is a
blend of #86 and #211 and is indeed (in our opinion) more powerful. We
will post this soon.

--

Jeff Hobbs, The Tcl Guy, http://www.activestate.com/

Eckhard Lehmann

2006-09-29, 4:08 am

Jeff Hobbs wrote:

> TIP #86 had some flaws. We (ActiveState) are working on one that is a
> blend of #86 and #211 and is indeed (in our opinion) more powerful. We
> will post this soon.

Thank you. Again my impudent question ;-)... Is this going to be in
8.5?

Eckhard

Donal K. Fellows

2006-09-29, 7:03 pm

Eckhard Lehmann wrote:
> Thank you. Again my impudent question ;-)... Is this going to be in
> 8.5?

Given what everyone's said so far, the answer has to be "not in its
current form" with the understanding that this doesn't mean very much if
the current form is about to change.

Thanks for the evaluation. That sort of thing is tremendously useful.

Donal.

Jeff Hobbs

2006-09-29, 7:03 pm

Eckhard Lehmann wrote:
> Jeff Hobbs wrote:
>
>
> Thank you. Again my impudent question ;-)... Is this going to be in
> 8.5?

There are no guarantees, but the TIP would target 8.5. It is in final
review amongst a few people right now.

--

Jeff Hobbs, The Tcl Guy, http://www.activestate.com/

Sponsored Links