Chip 2001 Mobile

home *** CD-ROM | disk | FTP | other *** search

/ Chip 2001 Mobile / Chip_Mobile_2001.iso / palm / system / _palmemu / palmemu.exe / Docs / _News.txt < prev next >

Wrap

Text File | 2000-08-14 | 10.4 KB | 235 lines

======================================================================== News (latest changes) for the Palm OS Emulator Copyright (c) 1998-2000 Palm, Inc. or its subsidiaries. All rights reserved. Please send bug reports, comments, suggestions, etc. to devsupp@palm.com ======================================================================== ======================================================================== REVISION HISTORY Starting in 2.1d25 (and backdated to 2.1d24) I'm giving credit to the person or people responsible for convincing me to fix a bug or add a feature, in some cases even providing the source code for the bug or feature. These people's names appear in []'s. See the file Credits.txt for a list of people who have submitted actual changes to the project. ======================================================================== Changes for 3.0a7 (8/7/00) -------------------------- * Added m100 support. * Added Palm VIIx support. * Implemented the following previously defined but unimplemenated HostControl functions: HostMalloc, HostRealloc, HostFree, HostRename, HostRemove, and HostTmpNam. * Added the following HostControl functions: HostStat, HostTruncate, HostMkDir, HostRmDir, HostOpenDir, HostReadDir, HostCloseDir, HostAscTime, HostClock, HostCTime, HostGMTime, HostLocalTime, HostStrFPrint, HostTime, and HostUTime. * Renamed a bunch of HostControl types to follow Palm's naming convention of having types end in "Type". Added typedefs for backward compatibility with applications using the old type names. * When a session is reset, all HostControl resources are reclaimed. That is, any blocks allocated with HostMalloc are freed, any files opened with HostFOpen are closed, any directories opened with HostOpenDir are closed, and any pointers to returned string buffers (as with HostGetEnv) are now invalid and the buffers are inaccessible. * After creating a handle with MemHandleNew, locking it, remembering the pointer, and then freeing the handle with MemHandleFree, the application could write to the freed block using the now-stale pointer. Fixed. [Neil Rhodes] * Support TsmGlueGetFepGlobals accessing system globals. * Updated URLs to point to www.palmos.com. * Fixed double-dispose bug when handing the getting or setting of a linger-related socket option. The symptom of this bug is that Poser would crash when calling NetLibSocketOptionGet(...netSocketOptSock- Linger...). [Peter Burka] * Fixed problem in NetLibSendPB when NetLib redirection is turned on; the various parameters weren't converted correctly into something the host TCP stack could use. [Rob Leslie] * (Unix) Added dynamic check for socklen_t in configure.in. [Radu Cornea] * Internal change: added new support classes: EmPoint, EmRect, EmRegion, EmRefCounted, EmRefCounter, and EmPixMap. * Used the preceding classes in rewriting the pipeline that converts the LCD framebuffer into something that can be displayed on the host screen. These changes (a) better insulate much of Poser from the specifics of the LCD framebuffer format, (b) reduce the amount of platform-specific code needed to create a bitmap that can be blitted to the host screen, and (c) result in slightly faster emulation. * Poser makes use of the following as a compile-time assert: char _dummy[expr ? 1 : 0]; If (expr) is true, then the compiler evaluates the stuff inside the brackets as 1. Otherwise, the compiler evaluates it to zero, resulting in the invalid definition of a zero-sized array. However, unbeknownst to this poor programmer, gcc allows the definition of zero-sized arrays as an extension. Therefore, all instances of the above technique have been changed to COMPILE_TIME_ASSERT(expr), where COMPILE_TIME_ASSERT is a macro that expands to something like: char _dummy[expr ? 1 : -1]; Even gcc complains about that one. [Frank Yellin] * Now that we have a working compile-time assert, a number of errors popped up. Part of the NetLib redirection facility involves translating between NetLib data structures and host TCP (sockets) data structures. The process of translating back and forth assumed that various structs and constants between the two systems were the same and could be copied back and forth without any interpretation. This assumption was backed up by compile-time asserts. But those asserts were broken. Now that they're working, it was found that many assumptions were bad and/or wrong. Therefore, much of the translation code was rewritten to remove those assumptions. [Frank Yellin] * (Unix) Turn on -fexceptions unconditionally. While this is on by default for C++, we now turn it on for C, too. Poser seems to need this on Solaris/Sparc systems. [Frank Yellin] * (Unix) Implemented Save Screen... menu item. Files are saved to .ppm format. * (Internal) Implemented EmDlg::DoGetFile, DoGetFileList, DoPutFile, and DoGetDirectory on all platforms, and use them universally. * Added HostGetFile, HostPutFile, and HostGetDirectory. * (Unix) Use new "FileChooser" widget from Easy Software. This one will probably get incorporated into FLTK 2.0, and it just plain looks and works nicer anyway. [Easy Software] * Patched TblHandleEvent to fix a Palm OS 3.5 bug where the drawing state would eventually become invalid, resulting in incorrect drawing and possible ErrDisplays from the OS. [Tom Bulatewicz] * Inhibit error checking in NetLibBitMove. This function sometimes accesses bytes past the end of an allocate block of memory, resulting in a Poser error message, usually while using PQAs. As it turns out, it does this in a fairly safe fashion, so prevent the error message from appearing. * (Internal) Use new method to access low-memory Palm OS System globals -- one based on the EmPalmStructs mechanism. * The progress dialog that's displayed when installing files has changed. The progress indicator is now modelled after the Macintosh file copy progress dialog; instead of having two indicators -- one for files and one for the current file -- the single progress indicator now indicates overall progress for all files being installed. Also, the Stop button now works. Finally, the progress dialog is now also implemented on Unix. * Enabled logging of Exchange Manager actions (just calls to ExgSend and ExgReceive). * (Internal) New Session dialog is now implemented using cross- platform techniques that Poser is slowly moving towards. This change results in consistant behavior of the dialog across platforms. * (Unix) "configure" script now supports "--with-fltk=DIR" command line option. [Arturo Tena] * Poser contains facilities for extracting the "Macsbug" symbols that can be appended to compiled functions for debugging purposes. It finds these symbols using heuristic methods that don't always succeed. If they fail, fail gracefully. [Adam Dingle] * Fixed problem where turning off the "power" to a Palm IIIc would blank the LCD, but turning the "power" back on would not redraw the LCD. [Bob Petersen] * Fixed implementation of Perl wrapper for HostImportFile and HostExportFile. [Derek Johnson, Andy Ihlenfeldt, John Ludwig] * Poser would crash if you passed a full path to HostProfileDump instead of just the name of a file. * Ensure that the binary file produced by profiling an application ends with .mwp. * Include keyboard modifiers when posting key events from the host computer's keyboard. [Ken Krugler] * Changed "New Session" dialog to use MRU list for ROM Files, and not present devices for selection that cannot possibly run the chosen ROM File. * Updated CodeWarrior projects to CodeWarrior 5.3 (IDE 4.0.4). * Fixed problem with lack of responsiveness in mouse-click handling. The problem was that there was only a 1-element queue, as it were, for handling mouse-clicks. It was too easy for the user to click the mouse and let up on the button before the emulated Palm OS had a chance to even recognize that the mouse button was down. Now there is an n-element queue to handle mouse ups/downs. I had originally stayed away from that because I thought that it would be too easy for the user to overrun the event queue, zooming the mouse around faster than the Palm OS could keep up. But we'll see how this works out. Note that the hard buttons across the bottom of the the emulated device can still be unresponsive. That's because there's no queuing involved; the mechanism for handling them is different from that used to handle events in the LCD area, and can't easily be converted to do so. * (Unix) Fixed problem with specifying desktop serial port to use for ROM Transfer. [Rick Richardson] * Catch a whole class of problems where Poser didn't detect when the emulated program counter went bad. [Eugene Chin] * Fixed byteswapping problems with creating the binary output file for profiling. Such a file created on Windows couldn't be read into MW Profiler on a Mac. [Greg Simon] * (Internal) Rewrote all of the emulation code that deals with hardware registers (such as the Dragonball registers). There were many subsystems that had to deal with memory in this way (the Dragonball (328, EZ) register handlers, the SED register handlers, the Visor USB register handlers, the Palm VII EZ PLD register handlers, the Sybol ASIC register handlers, etc.). All of these subsystems originally were just cloned from each other and tweaked. This lead to a lot of duplicated code. All that's now been changed so that the common code is in EmBankRegs, and the hardware-specific code is in subclasses of EmRegs. * (Mac) Fixed crash that would happen when clicking on the mailto URL in the Palm section of the About box. * The Field Manager in Palm OS 1.0 - 3.2 has a bug that causes the message "Invalid insertion point location" to be erroneously displayed. Added a patch to ErrDisplayFileLineMsg to squelch this message if it looks incorrect. [Stephen P. Hill, David Kendall] * (Mac) Updated to GUSI 2.1.1, fixing a bug in HostFOpen, where creating the file ":x:y" would create a file called "x" if the directory didn't exist. * (Unix) Re-arranged checks at bottom of EmulatorTypes.h to allow for definition of INADDR_LOOPBACK. [Jon Fo] * (Unix) Specify -fpermissive to gcc to support broken X11 headers. [Jon Fo]