diff options
Diffstat (limited to 'tcl/mac/bugs.doc')
-rw-r--r-- | tcl/mac/bugs.doc | 86 |
1 files changed, 38 insertions, 48 deletions
diff --git a/tcl/mac/bugs.doc b/tcl/mac/bugs.doc index 120489d9fbd..a4936e2e04f 100644 --- a/tcl/mac/bugs.doc +++ b/tcl/mac/bugs.doc @@ -1,54 +1,44 @@ -Known bug list for Tk 8.0 for Macintosh +Known bug list for Tcl 8.0 for Macintosh -Originally by Ray Johnson +by Ray Johnson Sun Microsystems Laboratories rjohnson@eng.sun.com -Maintained by: -Jim Ingham -Cygnus Solutions, a Red Hat Company -jingham@cygnus.com - RCS: @(#) $Id$ -We are now very close to passing the test suite for Tk. We are very -interested in finding remaining bugs that still linger. Please let us -know (and send us test cases) of any bugs you find. - -Known bugs: - -* Transient windows (set by wm transient) do not go away when the - master does. - -* Tearoff menus should be floating windows & floating windows should - float. They also shouldn't be resizable. - -* The -use and -container windows only work with other Tk windows in - the same process. Also, if you try really hard (for instance by binding - on Destroy of an embedded window and destroying the container's toplevel) - you can get Tk to crash. This should never be necessary, however, since - the destruction of the embedded window triggers the destruction of the - container, so you can watch that instead. - All the focus bugs in Tk8.0 have been fixed, however. - -* The send command is only implemented within the same app. - -* You cannot color buttons, and the indicators for radiobuttons and - checkbuttons under Appearance. They will always use the current - Theme color. But, then, you are not supposed to... - -* Drawing is not really correct. This shows up mostly in the canvas - when line widths are greater than one. Unfortunantly, this will not - be easy to fix. - -* The active menu highlight color in Tearoff menus will not match the system-wide - menu highlight color under Appearance. It will be black instead. This is not - easy to fix, since the Appearance API's don't really allow you to get your hands - on this information... - -There are many other bugs. However, will no get listed until they -are reported at least once. Send those bug reports in! - - - -Jim +This was a new feature as of Tcl7.6b1 and as such I'll started with +a clean slate. I currently know of no reproducable bugs. I often +get vague reports - but nothing I've been able to confirm. Let +me know what bugs you find! + +The Macintosh version of Tcl passes most all tests in the Tcl +test suite. Slower Macs may fail some tests in event.test whose +timing constraints are too tight. If other tests fail please report +them. + +Ray + +Known bugs in the current release. + +* With the socket code you can't use the "localhost" host name. This + is actually a known bug in Apple's MacTcp stack. However, you can + use [info hostname] whereever you would have used "localhost" to + achive the same effect. + +* Most socket bugs have been fixed. We do have a couple of test cases + that will hang the Mac, however, and we are still working on them. + If you find additional test cases that show crashes please let us + know! + +* In Tcl 8.2, the new Regexp code seems to be more deeply recursive than +the older version in Tcl8.0. As a result, I have had to increase the Stack +size of Tcl to 1Meg. If you are not doing regexps with many subexpressions, +this is probably more stack than you will need. You can relink with the +stack set to 512K, and you will be fine for most purposes. +* This regexp problem is fixed in Tcl8.3. If you are going to do complex +regexp's, it is probably a good idea to keep the stack size big. But normal +regexps will not cause crashes. + +* The "clock scan -base" command does not work. The epoch is wrong. +* The file mtime command does not work when setting the time, it is off +by 4 years. |