gmane.comp.lang.tcl.core

Re: please vote tip 312

Donal K. Fellows — 2008-08-29T21:49:55

By our build system? Surely you jest! None of them ever were AIUI, but the big-iron folks tend to be used to hacking Makefiles. (To be honest, I wouldn't know about this except I've fielded bug reports on the topic during the 8.4 development cycle. In 8.5, we've not done anything that would tickle the problems; we're now clean about INT2PTR/PTR2INT macro usage.) Donal. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: please vote tip 312

Jeff Hobbs — 2008-08-29T21:30:34

Er, but can you name one that has actually been supported by Tcl's build system in the last several years? Jeff ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: Coroutines in HEAD

Alexandre Ferrieux — 2008-08-29T15:42:22

Again, a followup would be appreciated... On Fri, Aug 29, 2008 at 1:23 AM, Alexandre Ferrieux public.gmane.org> wrote: ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: Coroutines in HEAD

Donal K. Fellows — 2008-08-29T14:45:19

It also comes unstuck with local variable unset traces. Probably other things too. Donal. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: "finally" support

Twylite — 2008-08-29T13:27:31

Thanks Donal, I have a working (I think) try/catch implementation in Tcl, but now that I've read TIP#90 in more detail I think I could improve it and address some of the hazy behaviour questions. I've looked at the [dict for] source, and I'm afraid I don't see how it applies. In the case of [dict for] you issue one command (dict, subcommand for) which, before it returns, cleans up resources. As I understand it Tcl_DictObjFirst() locks the dict internal rep (if the dict is not empty) which must later be unlocked (implicitly by Tcl_DictObjNext() on the last element or explicitly by Tcl_DictObjDone() ). The implementation ensures that Tcl_DictObjDone() is called before returning the result (TCL_OK, TCL_ERROR, ...) and resultObj. In other words all resource allocation and finalisation is within the scope of one call. My proposal for "finally" sees [finally] as its own command that registers a callback, and the callback is invoked at the end of the enclosing scope (a [try], a [catch], a proc/met

Re: Feature freeze coming, looking for TIP sponsors

Donal K. Fellows — 2008-08-29T13:09:58

Do you want me to change the title of this TIP for you? It's not really accurate any more... Donal. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: Coroutines in HEAD

Donal K. Fellows — 2008-08-29T13:04:39

It doesn't appear to do so at all with what you've proposed. ;-) You can't do that unless you have the state be entirely explicit. If you are relying on some critical piece of state that can only exist in the non-string representation, you can't go to another thread because inter-thread messaging (other than via the shared variable primitives of the Thread extension) is strictly string-based. Either you'll need to do much more magic (and locks and ...) or you'll not have suspend/resume that work in the way we currently understand them to. In that case you are using a non-Tcl state/value model. The Tcl value model is that of immutable transparent strings; you've got what you've got (it doesn't change under your feet), you can see what you've got (you can look inside), and do anything you like with it (e.g. you can stuff it over a channel trivially). Things that cannot meet that model are not Tcl values, and must be dealt with through named handles. Those powers seem (without studying anything very hard

Feature freeze coming, looking for TIP sponsors

Peter Spjuth — 2008-08-29T12:46:45

------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ Tcl-Core mailing list Tcl-Core-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f< at >public.gmane.org https://lists.sourceforge.net/lists/listinfo/tcl-core

Re: Coroutines in HEAD

Lars Hellström — 2008-08-29T12:38:02

Neil Madden skrev: FOTC. To "pass the command name" is usually considered to be pass-by-reference. If that sufficies for being "first-class", then arrays are "first-class" too! You mean, like a [foreach] where list elements would be generated as they are needed? Roughly as follows: proc TCL_SUSPEND_handler {bgerror cont opt} { if {[dict get $opt -code] != $::TCL_SUSPEND} then { return [{*}$bgerror $cont $opt] # Or [tailcall], if you prefer that. } # Now for the TCL_SUSPEND handling... switch -- [lindex $cont 0 0] "after" { # Syntax: suspend after $ms after [lindex $cont 0 1] [list resume $cont] } "vwait" { # Syntax: suspend vwait $varname trace add variable [lindex $cont 0 1] write [list ::apply { {varname cont name1 name2 op} { # Trace removal omitted after 0 [list resume $cont] } :: } $varname $cont] } "call-cc" { # Syntax: suspend call-cc $cmd ?$arg ...? # The $cmd

Re: please vote tip 312

Donal K. Fellows — 2008-08-29T09:32:59

Unfortunately, that's not true. Some platforms (especially the more exotic types of big-iron) have the size of char* being different from the size of void*! Really. Hence, while I was planning to do something about this, I wasn't going to touch it until we start working on 9.0. (There are many other places where we are using the wrong types in our C API, such as the argument to Tcl_Alloc. I look forward to fixing them. But absolutely not during 8.*, since they'd destroy ABI compatability.) Definitely! Don't add new mistakes. :-) Donal. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: Coroutines in HEAD

Donal K. Fellows — 2008-08-29T09:24:23

This is only valuable when you are running a workload where the service cost of a single session varies widely. If the cost of serving a single client is fairly uniform, you get good enough load balancing from just using a thread pool. The down-side of relocating things across threads is that it makes memory management *much* nastier, and introduces a whole bunch of hot global locks. Don't go there, please! Donal. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: "finally" support

Donal K. Fellows — 2008-08-29T09:18:57

I think so. The main issue from my perspective is how to handle "catch" clause matching. Work that out and the rest is trivial enough. (The other big issue is that Tcl's not very good about its errorCode production, but before now there's not been much incentive to change...) It's pretty easy actually, since there's already code that needs that sort of thing behind the scenes. Look at how [dict for] works for inspiration, as that needs "finally" handling to drop the lock on the internal representation of the dictionary being iterated over. At the Tcl level, take a really good look at the facilities introduced by TIP#90 as they make everything much simpler. Should be easy in an extension if no compilation to bytecode is required. With the application of a bit of TIP#90 magic, you should be able to do this in pure script. The performance boost from a C or bytecode implementation would be considerable though. Donal. ------------------------------------------------------------------------- This SF.Net

Re: please vote tip 312

Donal K. Fellows — 2008-08-29T09:10:13

We don't need to perpetuate the mistakes of the past. :-) Donal. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: please vote tip 312

Jan Nijtmans — 2008-08-29T08:20:34

2008/8/28 Rene Zaumseil public.gmane.org>: Just a small remark: int Tcl_LinkArray(Tcl_Interp *interp, CONST char *varName, char *addr, int type, int size) shoulnd'nt that be: int Tcl_LinkArray(Tcl_Interp *interp, CONST char *varName, void *addr, int type, int size) because 'addr' could point to anything and we don't want to cast to (char *) everywhere Further on, looks good. Regards, Jan Nijtmans ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: please vote tip 312

Jan Nijtmans — 2008-08-29T08:52:28

2008/8/29 Rene Zaumseil public.gmane.org>: Well, there are a lot of functions that would better fit a void * than a char * argument or return value. e.g. char * Tcl_Alloc (unsigned int size) --> void * Tcl_Alloc (unsigned int size) void Tcl_Free (char * ptr) --> void Tcl_Free (void * ptr) (and all variations.....) So now we know one more: Tcl_LinkVar (Tcl_Interp *interp, CONST char *varName, char *addr, int type); --> Tcl_LinkVar (Tcl_Interp *interp, CONST char *varName, void *addr, int type); I think this is worth a new TIP (unrelated to TIP #312, of course). It wouldn't give any problems because there is zero risk for incompatibilities. But better do it 'right' for new API's. Regards, Jan Nijtmans ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip fo

"finally" support

Twylite — 2008-08-29T08:42:02

Hi everyone, I've been mulling this over for some time, and DGP's announcement of an impending feature freeze means that it's time I ask about this now, or leave it for 8.7. Tcl's lack of a standard syntax for try/catch or try/finally is ... a problem. There are extensions that provide such support of course, and there is TIP 89 (Try/Catch Exception Handling in the Core http://www.tcl.tk/cgi-bin/tct/tip/89) which seems to be doing nowhere. I'd like to make that case that (1) we should try and get a try/catch syntax into the core ASAP, and (2) we should consider an alternative "finally" syntax that is _better_ than the standard structured try/catch/finally. In Tcl I see this idiom a lot: - set handle {} set failed [catch { ... unsafe operations ... } em opts] ... finally clauses ... if { $handle ne {} } { ... free resource ... } if { $failed } { ... catch/error handler, perhaps log and rethrow the error ... } - 1. There is a bunch of "dead syntax" in there - we're wasting space trying

Re: please vote tip 312

Rene Zaumseil — 2008-08-29T08:28:03

Am Freitag, 29. August 2008 10:20 schrieb Jan Nijtmans: I have no objections. But the char* is just the same as in the existing Tcl_LinkVar() function. rene ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: Coroutines in HEAD

Neil Madden — 2008-08-29T14:06:36

Flight of the Conchords? Pass by value/reference is orthogonal to whether something is first- class. E.g. Java objects are considered first class, but are passed by reference. You can indeed have "first class references" (see e.g. ML), without that being an oxymoron. The difference between commands and arrays is that you don't have to perform any extra step to use a passed command name, i.e.: proc foo {cmd args} { $cmd ... } vs proc foo {arrName args} { upvar 1 $arrName arr; something $arr (foo) .. } Essentially yes. Python's [for] statement can operate in this manner. OK, so you tear down the entire stack and then recreate it again? Or do you just call the command from the toplevel? If the latter, that's not call/cc. It's an intriguing idea, but I have to wonder - why bother? This worries me quite a lot, actually. People almost certainly would start relying on these implementation details. Non-determinism can also be simulated with coroutines -- this is one of the examples

Re: Coroutines in HEAD

miguel sofer — 2008-08-28T23:37:56

Nope - if someone wanted to make things work according to that scenario, my coroutines will not support her. It would be "Overly Cool" but it does not seem likely to happen anytime soon. ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: Coroutines in HEAD

Alexandre Ferrieux — 2008-08-28T23:28:23

Oh, so you want to pack a full coro's context (including an army of globals) into serialized form and make it migrate to the random thread chosen, instead of just waking up the proper thread after a simple lookup ? Either you're kidding or I'm missing something elephant-size... -Alex ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Re: Coroutines in HEAD

Alexandre Ferrieux — 2008-08-28T23:23:26

Indeed. See: http://groups.google.com/group/comp.lang.tcl/tree/browse_frm/thread/7244a14c3a82b090/cd94a2dff6910b36?hl=en&rnum=1&q=in+head&_done=%2Fgroup%2Fcomp.lang.tcl%2Fbrowse_frm%2Fthread%2F7244a14c3a82b090%2F2361d9dc487bd8d9%3Fhl%3Den%26lnk%3Dgst%26q%3Din%2Bhead%26#doc_4244d504a9b42828 Yeah, I'm impatient to know what. That is true but does not shed light on the *need* for [suspend/resume] instead of [coroutine]: Classical event-driven style imposes non-contiguous code, while both [coroutine] and your approach allow for straightforward style -- this doesn't break the tie. So, if you have no personal interest in serialized continuations *and* just want the linear style that [coroutine] already provides, what's the point ? -Alex ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open S