gmane.comp.lang.tcl.core

why not USE_TCLALLOC in unix??

miguel sofer — 2008-09-06T19:16:41

Why are we *disabling* our own pooled allocator in unix non-threaded configurations? See unix/Makefile.in line 985: it is disabled even as a configure-time option, it can only be enabled by editing the makefile How bad would it be to start using it by default (as in windows)? The switch to disabling the system's malloc by default was done on 1999-03-10 by stanton, the CL comment does not address the issue (it says "Merged stubs changes into mainline for 8.0"): see line 645 in http://tcl.cvs.sourceforge.net/tcl/tcl/unix/Makefile.in?r1=1.17&r2=1.18 I'm planning to default to USE_TCLALLOC also on unix non-threaded builds. Anybody against? AFAIR, the arguments are: * sys malloc is slower, but mostly "fast enough" * users complain about the high-water-mark allocator ---- The reason I would like to do this: I want to get rid of the Tcl execution stack altogether, and alloc CallFrames and related structs directly. Pros: * faster than sys malloc on unthreaded builds * remove delicate and somewhat duplicated code (Tcl stack) * kill off TclStackAlloc with its restrictions; replace with equally fast alternative TclInterpAlloc(interp, size) * first step towards enabling CallFrames to live in a tree-not-stack * first step towards making many allocs faster in the threaded build: pass interp to the alloc/free functions when it is available. Ie, replace many ckalloc/Tcl_Alloc calls with TclInterpAlloc Cons: * slow when using system malloc I know how to do this at zero runtime cost with either zippy or our own pooling allocator (tclAlloc.c): zippy needs a small modif to use interp's allocCache field instead of looking up the thread's cache. But I do not think it'd be a good idea to make the default unthreaded build slower. Side note: Eventually I would like to merge the code in tclAlloc.c and tclThreadAlloc.c - the first can be seen as a special case of the second (no locks, no shared blocks), we do not need this stuff twice in there. ------------------------------------------------------------------------- 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=/

[Fwd: Re: Tcl 9.0]

Donal K. Fellows — 2008-09-06T17:15:00

[My response to the previous] -------- Original Message -------- Subject: Re: Tcl 9.0 Date: Fri, 05 Sep 2008 21:41:18 +0100 From: Donal K. Fellows manchester.ac.uk> Organisation: University of Manchester To: Ron Zajac gmail.com> References: mail.gmail.com> Ron Zajac wrote: There's much that we can do without needing the 9.0 step. For example, we're in the process of getting a new implementation that supports coroutines and objects. That's going to be part of 8.6, which we hope will go into public beta at the Tcl Conference this fall. (The scripted object support is also available as an extension for 8.5 called TclOO; those good folks at ActiveState do a packaged version for all their supported platforms.) But we're unlikely to go with a value model like that used by Perl and Python. The issue is that our value model is basically that of immutable objects (we use copy-on-write for efficiency) and theirs is of mutable objects; we prefer to hide mutability behind immutable handles. This difference has both good and bad implications — we can actually manage memory very efficiently, but can't share engine implementations without major effort — but we're willing to put up with them. Well, there's a major reengineering of things going on right now (Miguel Sofer's Non-Recursive Execution Engine is roughly the Tcl equivalent to Stackless Python) but that's going in *before* the 9.0 step. When we do 9.0, we're more likely to be doing it as something of a housekeeping change. Or we might decide to try doing compiling to native code. We don't have a grand plan for this all yet. :-) Indeed, it's fair to say that we don't have a concrete plan for the 9.0 transition yet (beyond a few specific things relating to types in the C-level ABI). I'm in favour of keeping as many existing scripts working as possible, so long as they don't meddle in nasty places. An example to illustrate something that is a technical "bad change" but which won't hit almost everyone is the exact moment when the first word of a command call is bound to the implementation for call. Right now, that happens after all the other words of the call are substituted, but changing it so that it happens before the substitution of the arguments at all will make things quite a lot easier to compile, and almost no scripts will be broken by it (you'd need to be changing the implementation out from under a command while substituting one of the arguments to that command, which is rather nasty and tricky). Maybe going even further along that road would enable even more efficient execution without deep impacts; I'm just speculating right now, not designing. :-) I appreciate your concerns, and I feel that I should note that the wishlists on the wiki (and in our feature request databases on SourceForge) are just that: wishlists by lots of people in the community. One of the things that I (and my fellow Core Team folks) have to do is to figure out which of those things are possible to implement, a good idea, and able to mesh together to form a harmonious whole. Can I forward your message to the tcl-core mailing list? I think there are other people who should read it. 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=/ _______________________________________________ Tcl-Core mailing list Tcl-Core< at >lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/tcl-core

[Fwd: Tcl 9.0]

Donal K. Fellows — 2008-09-06T17:01:11

[Forwarded with permission] -------- Original Message -------- Subject: Tcl 9.0 Date: Thu, 4 Sep 2008 16:44:55 +0800 From: Ron Zajac public.gmane.org> To: donal.k.fellows-m7QheJJv02N4oUbgFh0ZNQ< at >public.gmane.org Mr. Fellows, First, thanks for your longstanding dedication to tcl; I consider myself (in my own humble way) to be a distinct beneficiary of the hard work of folks like yrself. Tcl's grown up to be a useful, easy-to-use, fun language, and I'm very grateful for all you've done. I was curious to see that people are starting to ruminate on 9.0. I had a glance at the Wishlist. My expectation was that I would see stuff that would meet or rival the 7 --> 8 changes. I recall that the biggest change, 'way back when, was the management of data as "smart," dynamically typed objects and the corresponding serious optimizations that catapulted tcl from a compact, distinctly "embeddable" language to an all-purpose scripting language that could then compete properly with Perl, python, and the other new-generation scripting languages that came about to fill the yawning gap between the cheap scripting of sh/bash/ksh/et.al . and full-bore high-level languages for "serious" development, like C++. What I'm seeing in the 9.0 considerations, while serious, high-level stuff, is nothing like that level of dramatic rethinking of core operations represented in the 7 --> 8 step. Now it may be that the current inner workings of tcl's precompile and runtime execution are beautiful, elegant, and pretty much seen as a satisfying pinnacle of fully evolved scripting language architecture. And I'm cool with that. But it seems to me that whatever you folks do as you start to develop a 8 --> 9 plan, you should keep the 7 --> 8 history in your sights. Is there no larger vision for a major rethinking of tcl's system operations, analogous to the 7 --> 8 step? If not, it may be a little misleading to bill a next version as "9.0." I think I understand part of the rationale for the major version step may involve a loss of complete backwards compatibility, therefore necessitating a step "up" to 9.0. While I seem to recall that the 8.0 abstraction of internal dynamic typing did leave a few sloppy programmers in the lurch, it recall the 7 --> 8 development sought to emulate the Perl 4 --> 5 evolution in terms of an attempt to make things as transparent as humanly possible to the Users "out there." A quick review of the "9.0 Wishlist" gives me the impression that the evolution to 9.0 will abandon this way of thinking; that the major version jump is supposed to communicate to the community that their 8.x code will (perhaps) likely not function under a future 9.0. Is that right? Anyway: I just wanted to drop these ideas. I'm an old-timer. And, I would be remiss not to add, I'm not a deeply technical User. For instance, I recently wrote a nicely functioning crossword puzzle development system using tcl (not tk!; it's tty!), and it's got clever, clever bits in there that make it quite workable (have a look at http://www.ronazajac.com/Xword/ ); nonetheless I didn't avail myself of "higher" functionality largely because I'm not really a very technical guy. I tell you all this to emphasize that my understanding of the upper stratospheres of tcl development pretty much elude me. But I *do* well recall the lovliness of the 7 --> 8 step and the great benefits that brought even to lower-level Users like yr hmbl svt. I guess I'm dropping my 2-bits re an expectation some Users might have that a major version step should have the same "resonance" as the previous one. Well! If you read this far, yr either patient, loony, or perhaps found my ideas curious. Whatever you think, I do want to make sure you're left with the impression that tcl has been *very* good to me. Thanks, again! -raz (Ron A. Zajac, aka 冉雍時), Taipei, Taiwan

Question on TK not finding fonts

Mark Barton — 2008-09-03T14:37:11

Hi All, I have googled this one to death and still have not found any solution and forgive me if this is not the right list to address the question. I want to use TCL/TK on a Gumstix Verdex (X-Scale) device which uses OpenEmbedded build tools. The current build environment from the Gumstix folks does have version 8.4.11 TCL/TK and it seems to be working correctly except for the font selection. When I run Wish and run the command ""font families", only fixed is returned. I do get a list of fonts (Bitstream Vera Sans and Serif among others) if I run fc-list. The X server is kdrive (tiny-x) which as I understand is a reduced footprint version of Xfree86. The window manager is Matchbox. So there must be a disconnect between TK and the X server. Could this be a configuration issue at build time or should I be looking at Tiny-X side of the house? Any help is appreciated. Mark ------------------------------------------------------------------------- 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=/

TIP #326: Add -group Option to lsort

Kieran Elby — 2008-09-03T12:53:45

TIP #326: ADD -GROUP OPTION TO LSORT ====================================== Version: $Revision: 1.2 $ Author: Kieran Elby State: Draft Type: Project Tcl-Version: 8.6 Vote: Pending Created: Monday, 01 September 2008 URL: http://purl.org/tcl/tip/326.html WebEdit: http://purl.org/tcl/tip/edit/326 Post-History: ------------------------------------------------------------------------- ABSTRACT ========== This TIP adds a new option, *-group*, to *lsort* to request that a list be treated as consisting of repeated groups of elements (as opposed to sublists), and that the the groups be sorted according to a chosen element within each group. RATIONALE =========== Flat name-value pair lists are common in Tcl - consider the output of *array get* and *dict create* or the input to *foreach*. It is surprising then that there is no command to directly sort such a list by either the name or the value elements, while preserving the name-value mapping. Doing so currently requires turning a name-value pair list into a list of sublists, using the *lsort* /-index/ option to sort them, then flattening the list again, which is rather fiddly and inefficient. While sorting name-value pair lists is no doubt the most common use case, lists containing groups of more than two elements are also reasonably common, and so providing support for groups of any size seems useful and no harder to implement. The option format follows the Feature Request here: PROPOSED CHANGE ================= A new option, *-group*, taking one parameter, /grpSize/ will be added to the *lsort* command. If *-group* is supplied, the list will be treated as consisting of groups of /grpSize/ elements, and the groups will be sorted by either their first element or, if the *-index* option is used, by the element within each group given by the first index passed to *-index* (which is then ignored by *-index*). Elements always remain in the same position within their group. The list length must be a multiple of /grpSize/, which in turn must be at least 2. EXAMPLES ========== lsort -group 2 {carrot 10 apple 50 banana 25} returns "apple 50 banana 25 carrot 10" lsort -group 2 -index 1 -integer {carrot 10 apple 50 banana 25} returns "carrot 10 banana 25 apple 50" lsort -group 3 -index {0 1} {{Bob Smith} 25 Audi {Jane Doe} 40 Ford} returns "{Jane Doe} 40 Ford {Bob Smith} 25 Audi" (since Smith, which is at index 1 of element 0 of the first group comes after Doe, which is at the same position in the next group of 3.) REFERENCE IMPLEMENTATION ========================== Available online at COMPATIBILITY =============== Since this is a new and unambiguous option, and since the distinction between the options and the list to sort is clear, no compatibility problems are expected. LIMITATIONS ============= Only one element in each group can be used for comparison; it's not possible to compare groups element-by-element. FURTHER THOUGHTS ================== There may be some merit in adding a similar option to *lsearch*. The name-value pair case is common enough that an argument could be made for adding *-pairname* and *-pairvalue* options as synonyms for *-group 2 -index 0* and *-group 2 -index 1* respectively. COPYRIGHT =========== This document has been placed in the public domain. ------------------------------------------------------------------------- TIP AutoGenerator - written by Donal K. Fellows ------------------------------------------------------------------------- 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=/

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

please vote tip 312

Rene Zaumseil — 2008-08-28T18:59:11

Hello, please call a vote for tip 312. Reference implementation could go in 8.6 and 8.5. Thank you 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=/

Tcl/Tk 8.6 feature freeze and b1 release schedule

Donald G Porter — 2008-08-27T21:00:59

My aim is to get Tcl/Tk 8.6b1 released in source code form in time for an 8.6b1-based ActiveTcl to be distributed at the conference. After discussing with the Activators, I've set the target release date as Friday, October 10, 2008. If there are TIPs or new-features-yet-to-be-TIPped that someone wants to argue need to be part of Tcl/Tk 8.6 and those features cannot be completed by October 10, now is the time to make the case that we should delay feature freeze. What should we wait for, and why, and how long? Unless someone speaks up on that point, I'll proceed on the Oct. 10 release plan for Tcl/Tk 8.6b1, and everyone with TIPs that they do think can be done by then should proceed getting them done by then. Please use TIP 311 to indicate your plans. Thanks all.

TIP #325: System Tray Access

David N. Welton — 2008-08-25T18:12:47

TIP #325: SYSTEM TRAY ACCESS ============================== Version: $Revision: 1.1 $ Author: David N. Welton State: Draft Type: Project Tcl-Version: 8.6 Vote: Pending Created: Monday, 25 August 2008 URL: http://purl.org/tcl/tip/325.html WebEdit: http://purl.org/tcl/tip/edit/325 Post-History: ------------------------------------------------------------------------- ABSTRACT ========== Modern operating systems have a "system tray", where programs may place an icon to indicate program status. This TIP proposes that Tk should adopt some existing code in order to permit cross platform access to this functionality. EXISTING CODE =============== * Windows: winico * Linux: tktray * Mac: ??? Existing code is sufficient, and utilizes an acceptable license in order to repurpose it for a systray command. INTERFACE =========== To be determined by the TCT, but along the lines of what tktray provides: *tktray::icon* /pathName/ ?/options/? Create a new icon for the system tray. The application managing the system tray is notified about the new icon. It normally results in the icon being added to the tray. If there is no system tray at the icon creation time, the icon will be invisible. When a new system tray appears, the icon will be added to it. Options: -class: WM_CLASS attribute for the icon window. Tray manager may use class name to remember icon position or other attributes. -image: image to show in the system tray. The value must be the name of a photo image. Transparency data of the photo are used to set the window's shape. The icon will be automatically redrawn or resized appropriately on any image modifications. -visible: boolean value indicating whether the icon must be visible. The system tray manager continues to manage the icon whether it is visible or not. Thus when invisible icon becomes visible, its position on the system tray is likely to remain the same. COPYRIGHT =========== This document has been placed in the public domain. ------------------------------------------------------------------------- TIP AutoGenerator - written by Donal K. Fellows ------------------------------------------------------------------------- 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/Tk 8.6a2 RELEASED

Donald G Porter — 2008-08-25T15:28:35

Tcl/Tk 8.6a2 Release Announcement August 25, 2008 The Tcl Core Team is pleased to announce the 8.6a2 releases of the Tcl dynamic language and the Tk toolkit. This is the second alpha release of Tcl/Tk 8.6. More details can be found below. We would like to express our gratitude to all those who submit bug reports and patches. This information is invaluable in enabling us to identify and eliminate problems in the core. Where to get the new releases: ------------------------------ Tcl/Tk 8.6a2 sources are freely available as open source from the Tcl Developer Xchange web site at: http://www.tcl.tk/software/tcltk/8.6.html This web page also contains additional information about the releases, including new features and notes about installing and compiling the releases. Sources are always available from the Tcl SourceForge project's file distribution area: http://sourceforge.net/project/showfiles.php?group_id=10894 Binaries for most major platforms are available from: http://www.activestate.com/Tcl For additional information: --------------------------- Please visit the Tcl Developer Xchange web site: http://www.tcl.tk/ This site contains a variety of information about Tcl/Tk in general, the core Tcl and Tk distributions, Tcl development tools, and much more. Summary of Changes since Tcl/Tk 8.6a1: -------------------------------------- The following were the main changes in Tcl/Tk 8.6a2. A complete list can be found in the changes file at the root of the source tree. The more complete ChangeLog is also included with each source release. This release is the second alpha release of 8.6. The alpha moniker means that it is in feature addition mode. All bug fixes (and some more) up to and including 8.5.4 changes are included in 8.6a2. The following list focuses on new features added so far in 8.6. This release is a development release, and should only be considered for deployment use after considerable testing. * [TIP 304] New command [chan pipe]. * New Non-Recursive Evaluation implementation decouples the Tcl evaluation stack from the C evaluation stack. * New experimental commands in ::tcl::unsupported : [tailcall], [atProcExit], [coroutine] and [yield]. * Stack overflow protection in regexp engine. * Tcl_Finalize() no longer implicitly called on DLL_PROCESS_DETACH. * Stopped providing workarounds for broken Itcl use of [namespace code]. Use recent enough Itcl releases that take care of themselves. *** POTENTIAL INCOMPATIBILITY *** * Revised public Tcl routines passing (Tcl_ObjType *), (Tcl_Timer *), and (Tcl_Filesystem *) arguments to use "const" to indicate read-only pointers. *** POTENTIAL INCOMPATIBILITY *** * Fixed broken handling of ***= regexps. * Fixed crash in [namespace inscope {}]. * [namespace import]ed command now fire execution traces. * More consistent "wrong # args" messages. * Revised number formatting in Tk scroll interface so that values no longer look like integers. -- Tcl Core Team and Maintainers Don Porter, Tcl Core Release Manager

Tcl/Tk 8.6a2 releases

Donald G Porter — 2008-08-22T20:05:04

The files making up the 8.6a2 releases of Tcl and Tk are now at: ftp://ftp.tcl.tk/pub/tcl/tcl8_6/ The usual release announcement will be sent out as soon as the release notes are written. By Monday at the latest.

Pre-CFV: TIP#320

Donal K. Fellows — 2008-08-22T09:59:26

This is a notice that I'm planning to do a CFV on TIP#320 soon. The TIP adds a mechanism for people to declare variables in classes and objects so that the variables are made available in each of the methods of those classes/objects without further explicit declaration. In other words, it makes using variables in classes much smoother. (They're still not "class variables" though; you'd make those using [upvar] during instance construction.) Do people think that the proposed API is right? Is having variables not isolated between superclasses and subclasses a good idea (I think so, since that makes for a system that feels "Tcl-ish" to me, rather than officious and fussy). Is not having subclasses pick up on their superclasses' variable declarations a good thing? (Again, I think so...) Now is the time to sort these questions out. :-) 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=/

Coroutines in HEAD

miguel — 2008-08-17T19:43:37

I managed to commit an experimental implementation of coroutines in time for 8.6a2. This provides two new commands ::tcl::unsupported::coroutine ::tcl::unsupported::yield A brief description can be found at http://msofer.com:8080/wiki?name=Coroutines, otherwise the code in tests/unsupported.test is the only documentation. Test! Enjoy! Help make it better Cheers Miguel ------------------------------------------------------------------------- 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/TK Mac Cocoa Port

J. Todd Slack — 2008-08-15T17:02:13

Hi All, I have been working on the TCL/TK Cocoa Port for a while now. What is the best wat to stay up to date with source releases? I was using 8.5.2 and 8.5.4 has been released. Is there a patch mechanism (or delta file) that I can take advantage of? I know that 8.6 is coming out as well as some point. Perhaps get the latest alpha and work with that? Carbon to Cocoa is an interesting port to take on. I must admit it is more relaxing that going to a day job everyday. Is Daniel Steffan on this list? Daniel you had promised me your notes from the Auqa port, etc. Thanks, -Jason ------------------------------------------------------------------------- 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/Tk 8.5.4 RELEASED

Donald G Porter — 2008-08-15T16:23:20

Tcl/Tk 8.5.4 Release Announcement August 15, 2008 The Tcl Core Team is pleased to announce the 8.5.4 releases of the Tcl dynamic language and the Tk toolkit. This is the fourth patch release of Tcl/Tk 8.5. More details can be found below. We would like to express our gratitude to all those who submit bug reports and patches. This information is invaluable in enabling us to identify and eliminate problems in the core. Where to get the new releases: ------------------------------ Tcl/Tk 8.5.4 sources are freely available as open source from the Tcl Developer Xchange web site at: http://www.tcl.tk/software/tcltk/8.5.html This web page also contains additional information about the releases, including new features and notes about installing and compiling the releases. Sources are always available from the Tcl SourceForge project's file distribution area: http://sourceforge.net/project/showfiles.php?group_id=10894 Binaries for most major platforms are available from: http://www.activestate.com/Tcl For additional information: --------------------------- Please visit the Tcl Developer Xchange web site: http://www.tcl.tk/ This site contains a variety of information about Tcl/Tk in general, the core Tcl and Tk distributions, Tcl development tools, and much more. Summary of Changes since Tcl/Tk 8.5.3: -------------------------------------- The following were the main changes in Tcl/Tk 8.5.4. A complete list can be found in the changes file at the root of the source tree. The more complete ChangeLog is also included with each source release. This is a patch release, so it primarily included bug fixes and corrections to erratic behavior. Below are only the most notable changes. * Restored ability to compile Tk against the latest X header files from X.org, which are incompatible with earlier Tk releases. * Restored ability to embed non-stub-enabled Tk in a C program as a static package using MSVC compilers, by providing a missing export. * Restored ability to find packages on Samba shares on Windows. * Fixed duplicate inode errors accessing native filesystem on Windows. * Fixed crash in [ttk::scale]. * Fixed crashes in unmapped toplevel windows. * Better internal rep preservation shimmering lists and dicts. * Improved Tcl Module maintenance with automatic [package provide]. * Improved portability of value returned by [winfo id]. * Fixed rare panic in [scan]. -- Tcl Core Team and Maintainers Don Porter, Tcl Core Release Manager

TCL 2008 Last Call for participation and papers

2008-08-14T04:59:42

***** TCL 2008 Last Call for participation and papers ****** Note extension of abstract deadline. http://www.tcl.tk/community/tcl2008/ On October 20-24, 2008, the city of Manassas, VA will play host to Tcl 2008. Join us for what promises to be a memorable, informative and fun conference. Tcl 2008 will be held at the Comfort Suites in Manassas, VA; about 30 miles outside of Washington DC . Tutorial sessions on various topics will be held October 20 and 21. The technical sessions will run October 22 through October 24. *** If you are a student check out the special student rates *** at the conference website: *** http://www.tcl.tk/community/tcl2008/ click Registration If you have an interesting Tcl paper to contribute, a tutorial to offer, a suggestion to give, submit it to the Tcl 2008 program commitee: "tcl2008 at tcl dot tk". Pretty much anything Tcl/Tk is acceptable. In the past we have had papers on: * Application of Tcl/Tk in industries as diverse as engineering, science, industrial controls, broadcasting, financial services, medical and electronic design * Networking with Tcl/Tk, including distributed applications and network management * Object frameworks for Tcl/Tk * New widgets and techniques for GUI design with Tk * Simulation and application steering with Tcl/Tk * Tcl/Tk on handheld and embedded devices * New Tcl extensions and add-ons, including Tcllib and Tklib * Tcl/Tk centric operating environments * Tcl/Tk in education and learning environments. This year is the first year that the Tcl community is participating in the Google Summer of Code. The conference program committee would like to encourage submissions that report on the Tcl projects selected for Google SOC 2008. Author Schedule: * Submit abstracts and tutorial proposals to Tcl2008 at Tcl dot tk by August 22, 2008 * Authors will be informed of acceptance by September 7 * Author materials should be submitted to the proceedings editor by October 15, 2008 This year we will be making the conference proceedings available as a published book. This may impact our plans for when author materials must be submitted. Watch future call for participation for more. Tcl 2008 provides an unparalleled opportunity to discuss your Tcl projects with some of the world's top Tcl experts. Past conferences have included notables like: * Jeff Hobbs: The Tcl Guy and ActiveState's point man for Tcl development and support. * Clif Flynt: Author of _Tcl/Tk:_A_Developer's_Guide_, as well as the popular TclTutor application. * Donal Fellows Author of TCL OO. * Miguel Sofer, Tcl Core team member and byte code compiler expert. * D. Richard Hipp, Tcl Core Team member and author of sqlite. * Mike Doyle: Director of Eolas. * Joe English, Author of the ttk toolkit formerly (and still) known as Tile. * Sean "Hypnotoad" Woods, author of 'that other' Tao. Confirmed Speakers: * Jeff Hobbs will present the annual ActiveState of Tcl talk. Conference Committee: Local Site: =========== Affiliation Sean Woods Test and Evaluation Solutions LLC Facilities chair Clif Flynt Noumena Corp. General Chair Program Committee: Sean Woods Test and Evaluations Solutions LLC Clif Flynt Noumena Corp. Steve Redler IV SR Technology Steve Landers Digital Smarties Cyndy Lilagan Iomas Research Kevin B. Kenny G.E. Global Research Center. Jeffrey Hobbs ActiveState Software Inc. Andreas Kupries ActiveState Software Inc. Ron Fox NSCL Michigan State University Donal Fellows University of Manchester Larry Virden Tcl FAQ Maintainer Mike Doyle Iomas Research Matthew M. Burke George Washington University Gerald Lester TicketSwitch USA LLC Richard Suchenwirth Siemens Industrial Solutions and Services Postal Automation Division Tcl 2008 would like to thank those who are sponsoring the conference: ActiveState Software Inc. Noumena Corp. SR Technology Tcl Association -- Sincerely, Andreas Kupries public.gmane.org> ------------------------------------------------------------------------------- ------------------------------------------------------------------------- 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/Tk 8.5.4 Release Notes

Donald G Porter — 2008-08-13T16:47:58

Available for review at: http://sourceforge.net/project/shownotes.php?group_id=10894&release_id=619603 Please send any additions or corrections ASAP, so they might be included before Friday's release.

porting to solaris11/x86

reshma — 2008-08-13T07:32:44

Tcl/Tk 8.5.3 Release Candidates & Schedule

Donald G Porter — 2008-08-08T19:02:04

I've posted the RC0 candidates for Tcl/Tk 8.5.4 at ftp://ftp.tcl.tk/pub/tcl/tcl8_5/ Please give them as thorough testing as you can. I'm working toward making the actual Tcl/Tk 8.5.4 release on Friday, August 15. I also expect to release Tcl/Tk 8.6a2 on August 15, or possibly a short time later, but don't expect to do RCs for that. If this plan for release schedule is a problem for anybody, please speak up and I'll revise it.

TIP 323: Do Nothing Gracefully

Donald G Porter — 2008-08-08T14:32:13

TIP #323: DO NOTHING GRACEFULLY ================================= Version: $Revision: 1.1 $ Author: Colin McCormack Don Porter Kevin B. Kenny State: Draft Type: Project Tcl-Version: 8.5 Vote: Pending Created: Wednesday, 06 August 2008 URL: http://www.tcl.tk/cgi-bin/tct/tip/323.html Post-History: ------------------------------------------------------------------------- ABSTRACT ========== A number of Tcl's built-in commands currently raise errors when given arguments that would cause them to do nothing. This proposal asks that they instead simply do nothing gracefully. RATIONALE =========== With the introduction of {*} syntax in Tcl 8.5 and its ability to convert one original word into zero words when an empty list is expanded, scripts using the syntax are forced to check for the empty list when passing expanded arguments to a command taking variable numbers of arguments, but arbitrarily rejecting zero arguments. For example, variable {*}$dict will currently fail if $dict has no entries. Letting no-ops simply be no-ops rather than errors greatly lightens the programming burden on callers. There may be some marginal performance benefits as well. Furthermore, there is no sound reason to constrain the argument count of these commands to be positive when an argument count of zero has a natural interpretation as a no-op. There is an elegance in permitting a function to range coherently over the complete domain. PROPOSAL ========== The syntax of the following commands shall be revised as described: 1. *file delete* - accept zero pathnames. 2. *file mkdir* - accept zero directories. 3. *global* - accept zero variable names. 4. *glob* - accept zero patterns, returning list of zero matching files (with *-nocomplain*) or "no matches" error. 5. *lassign* - accept zero variable names. [] 6. *linsert* - accept zero elements. [] 7. *lrepeat* - accept both /number/ = 0 and zero elements. [] 8. *my variable* - accept zero variable names. 9. *namespace upvar* - accept zero variable names. 10. *string first* - able to find empty string. [] 11. *string last* - able to find empty string. [] 12. *string replace* - change interpretation of /last/ < /first/ case to enable replacement of an empty substring anchored at /first/. 13. *tcl::tm::path add* - accept zero paths. 14. *tcl::tm::path remove* - accept zero paths. 15. *variable* - accept zero variable names. [] COMPATIBILTY ============== Most of the proposed changes convert errors into no-ops. There should be no compatibility concerns with those. The exceptions are the proposed changes to *string first*, *string last*, and *string replace*, where current behavior is so inexplicable, it would be called a bug if it weren't documented. Changing those commands is an incompatibility, but likely a welcome one. IMPLEMENTATION ================ When an implementation patch is completed, it will be logged in the SF Tracker and noted here. EXCLUSIONS ============ Some other candidates for this treatment are intentionally left out of this proposal where for various reasons their conversion is complicated in some way, or the benefits of conversion are not as clear cut. These other commands might still be revised in a similar way, but are left for other TIPs to accomplish it so that this TIP can focus only on the low-hanging fruit. 1. *after idle* 2. *dict exists* 3. *dict set d* /val/ (no /key/ argument) 4. *dict unset d* /val/ (no /key/ argument) 5. *dict update d* /script/ (no /key varName/ arguments) 6. *eval* 7. *expr* 8. *file join* 9. *interp eval* 10. *namespace eval* 11. *parray* [] This feature request is more about tolerating errors than about expanding acceptance over a complete input domain. Not in the spirit of this TIP. 12. *uplevel* 13. *upvar* - presence of the optional /level/ argument makes this one messy. COPYRIGHT =========== This document has been placed in the public domain. ------------------------------------------------------------------------- TIP AutoGenerator - written by Donal K. Fellows

TIP Editor?

Donald G Porter — 2008-08-06T15:21:43

Donal Fellows seems to be on vacation. Has he arranged with anyone else to take over TIP Editor duties in his absence? If no one speaks up, I'll step in and do that, as I have before. Unless Donal's selected someone else and they speak up, if you have new TIPs to publish, or need TIP editing beyond the reach of the online capability, please e-mail me.