Re: Bug #1 - (debug-disable 'debug) has no effect ...
Neil Jerram <[EMAIL PROTECTED]> writes: > >> Do we say that we don't want to fix `(array? car) => #t)' now because > >> Guile has behaved this way for a while? > > Marius> Yes, exactly. It wont be fixed in the stable branch. > > Sorry again; I didn't realize that we weren't planning to fix this in > the stable branch. Understanding this makes the overall picture > much clearer. Ahh, good. No need to say sorry. It only shows that we all do not indeed agree what the 1.6 branch is for. I will add an explanation to the branch, as per Thien-Thi's suggestion. ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
Re: Bug #1 - (debug-disable 'debug) has no effect ...
> "Marius" == Marius Vollmer <[EMAIL PROTECTED]> writes: Marius> Neil Jerram <[EMAIL PROTECTED]> writes: >> > "Marius" == Marius Vollmer <[EMAIL PROTECTED]> writes: >> Marius> Neil Jerram <[EMAIL PROTECTED]> writes: >> >> >> >> I'm not convinced. How can people be expecting debugging to be turned >> >> on after having explicitly said `(debug-disable 'debug)' in their >> >> .guile file? >> Marius> Because Guile behaved like that for quite a while. >> >> But it's a bug! Marius> Yes, and it is fixed in the non-stable branch. It's too late to fix Marius> it in the stable branch. Marius> It is too late because I have decided so. I know that this might not Marius> be popular, and it comes somewhat as a surprise. I don't do this Marius> because I like to be despotic, but because I think it is constructive. Marius> 1.6 has been too long in the making. Of course, there is a huge Marius> amount of improvements to be made in Guile, but that's why we have the Marius> unstable HEAD branch. OK, fair enough. Sorry for going on about this for so long. >> Do we say that we don't want to fix `(array? car) => #t)' now because >> Guile has behaved this way for a while? Marius> Yes, exactly. It wont be fixed in the stable branch. Sorry again; I didn't realize that we weren't planning to fix this in the stable branch. Understanding this makes the overall picture much clearer. Neil ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
Re: Bug #1 - (debug-disable 'debug) has no effect ...
> "Marius" == Marius Vollmer <[EMAIL PROTECTED]> writes: Marius> Neil Jerram <[EMAIL PROTECTED]> writes: >> > "Marius" == Marius Vollmer <[EMAIL PROTECTED]> writes: >> Marius> Neil Jerram <[EMAIL PROTECTED]> writes: >> >> >> >> I'm not convinced. How can people be expecting debugging to be turned >> >> on after having explicitly said `(debug-disable 'debug)' in their >> >> .guile file? >> Marius> Because Guile behaved like that for quite a while. >> >> But it's a bug! Marius> Yes, and it is fixed in the non-stable branch. It's too late to fix Marius> it in the stable branch. Marius> It is too late because I have decided so. I know that this might not Marius> be popular, and it comes somewhat as a surprise. I don't do this Marius> because I like to be despotic, but because I think it is constructive. Marius> 1.6 has been too long in the making. Of course, there is a huge Marius> amount of improvements to be made in Guile, but that's why we have the Marius> unstable HEAD branch. OK, fair enough. Sorry for going on about this for so long. >> Do we say that we don't want to fix `(array? car) => #t)' now because >> Guile has behaved this way for a while? Marius> Yes, exactly. It wont be fixed in the stable branch. Sorry again; I didn't realize that we weren't planning to fix this in the stable branch. Understanding this makes the overall picture much clearer. Neil ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
Re: Bug #1 - (debug-disable 'debug) has no effect ...
Neil Jerram <[EMAIL PROTECTED]> writes: > > "Marius" == Marius Vollmer <[EMAIL PROTECTED]> writes: > > Marius> Neil Jerram <[EMAIL PROTECTED]> writes: > >> > >> I'm not convinced. How can people be expecting debugging to be turned > >> on after having explicitly said `(debug-disable 'debug)' in their > >> .guile file? > > Marius> Because Guile behaved like that for quite a while. > > But it's a bug! Yes, and it is fixed in the non-stable branch. It's too late to fix it in the stable branch. It is too late because I have decided so. I know that this might not be popular, and it comes somewhat as a surprise. I don't do this because I like to be despotic, but because I think it is constructive. 1.6 has been too long in the making. Of course, there is a huge amount of improvements to be made in Guile, but that's why we have the unstable HEAD branch. > Do we say that we don't want to fix `(array? car) => #t)' now because > Guile has behaved this way for a while? Yes, exactly. It wont be fixed in the stable branch. You have say what branch you are talking about. We can certainly attempt to fix it in HEAD, as soon as possible. The fix can be very experimental, even, and we can let it ripe a bit. But we should not do this in the stable branch. Now, the 'debug-disable' fix might be considered simple but since it touches the public behavior of Guile in a common situation, we should just be conservative, even if the existing behavior is considered a bug. > Plus, the behaviour is going to change in 1.8, because someone has > already made a fix in HEAD. Given that it will change, surely it's > better to do sooner. We're not in the middle of a stable release > series yet. We are, effectively. And, the behavior might change again in HEAD until 1.8 comes out. ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
Re: Bug #1 - (debug-disable 'debug) has no effect ...
> "Marius" == Marius Vollmer <[EMAIL PROTECTED]> writes: Marius> Neil Jerram <[EMAIL PROTECTED]> writes: >> >> I'm not convinced. How can people be expecting debugging to be turned >> on after having explicitly said `(debug-disable 'debug)' in their >> .guile file? Marius> Because Guile behaved like that for quite a while. But it's a bug! Do we say that we don't want to fix `(array? car) => #t)' now because Guile has behaved this way for a while? Plus, the behaviour is going to change in 1.8, because someone has already made a fix in HEAD. Given that it will change, surely it's better to do sooner. We're not in the middle of a stable release series yet. Neil ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
Re: Bug #1 - (debug-disable 'debug) has no effect ...
Neil Jerram <[EMAIL PROTECTED]> writes: > > "Marius" == Marius Vollmer <[EMAIL PROTECTED]> writes: > > Marius> Neil Jerram <[EMAIL PROTECTED]> writes: > >> This bug has been fixed in unstable CVS, by removing the lines at the > >> end of (ice-9 debug) that unconditionally enable debugging. Is there > >> any reason why we shouldn't fix it in the same way on the stable > >> branch? > > Marius> This change would be too far reaching for a supposedly > Marius> stable program. Debugging wouldn't be turned on when > Marius> people expect it. > > I'm not convinced. How can people be expecting debugging to be turned > on after having explicitly said `(debug-disable 'debug)' in their > .guile file? Because Guile behaved like that for quite a while. Ok, I thought about my general view of the stable branch situation, and this came from it: We need to be much more rigorous about really keeping the stable branch stable. From now on, only release critical fixes will be allowed. I intend to be firm on this. We are not getting anywhere without restraining us from putting new stuff into the stable branch. I don't want to stop anyone from working on Guile, but I'd like to get 1.6 released fast. This will help all of us. See my posting to guile-devel. ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
Re: Bug #1 - (debug-disable 'debug) has no effect ...
> "thi" == Thien-Thi Nguyen <[EMAIL PROTECTED]> writes: thi>From: Neil Jerram <[EMAIL PROTECTED]> thi>Date: 08 Mar 2002 23:53:35 + thi>I'd say it's a bug, and that it would be better to squash it now than thi>to support strange expectations through 1.6.x. thi> it's also a bug in the sense that side-effects on load (except for thi> `define') are usually considered antithetical to clean modular design, thi> especially since the module system does not (and might never) support thi> "instantiable modules". this would be a good check to add to the thi> linter. thi> on the other hand, loading modules serves to extend the basic system in thi> other ways (e.g., syntax), so it might seem natural in a similar fashion thi> to piggyback turning on the debugging evaluator when loading (ice-9 thi> debug). A compromise fix would be to make (top-repl) autoload (ice-9 debug) rather than using it directly. Then the loading of (ice-9 debug) would turn on debugging the first time that someone uses `trace', which is quite intuitive. Neil ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
Re: Bug #1 - (debug-disable 'debug) has no effect ...
From: Neil Jerram <[EMAIL PROTECTED]> Date: 08 Mar 2002 23:53:35 + I'd say it's a bug, and that it would be better to squash it now than to support strange expectations through 1.6.x. it's also a bug in the sense that side-effects on load (except for `define') are usually considered antithetical to clean modular design, especially since the module system does not (and might never) support "instantiable modules". this would be a good check to add to the linter. on the other hand, loading modules serves to extend the basic system in other ways (e.g., syntax), so it might seem natural in a similar fashion to piggyback turning on the debugging evaluator when loading (ice-9 debug). in the long run, the stability of a bad design rarely outweighs the temporary instability that changing it to a good design entails, in terms of both user and maintainer satisfaction. (however, there are always exceptions; see Unix. ;-) thi ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
Re: Bug #1 - (debug-disable 'debug) has no effect ...
> "Marius" == Marius Vollmer <[EMAIL PROTECTED]> writes: Marius> Neil Jerram <[EMAIL PROTECTED]> writes: >> This bug has been fixed in unstable CVS, by removing the lines at the >> end of (ice-9 debug) that unconditionally enable debugging. Is there >> any reason why we shouldn't fix it in the same way on the stable >> branch? Marius> This change would be too far reaching for a supposedly Marius> stable program. Debugging wouldn't be turned on when Marius> people expect it. I'm not convinced. How can people be expecting debugging to be turned on after having explicitly said `(debug-disable 'debug)' in their .guile file? Also note that the fix only affects REPL usage, not programs run using `guile -s'. If you are concerned about `trace' not working, we could add `(debug-disable 'debug)' into the definition of `trace'. I'd say it's a bug, and that it would be better to squash it now than to support strange expectations through 1.6.x. Neil ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
Re: Bug #1 - (debug-disable 'debug) has no effect ...
Neil Jerram <[EMAIL PROTECTED]> writes: > This bug has been fixed in unstable CVS, by removing the lines at the > end of (ice-9 debug) that unconditionally enable debugging. Is there > any reason why we shouldn't fix it in the same way on the stable > branch? This change would be too far reaching for a supposedly stable program. Debugging wouldn't be turned on when people expect it. ___ Bug-guile mailing list [EMAIL PROTECTED] http://mail.gnu.org/mailman/listinfo/bug-guile
