MacFormat 1996 February

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

/ MacFormat 1996 February / macformat-034.iso / mac / Shareware City / Extensions / Mt. Everything Folder / History.p next >

Wrap

Text File | 1995-11-19 | 26.8 KB | 564 lines | [TEXT/ttxt]

************************************************************** ** Project: ** Mt.Everything cdev ************************************************************** Font Monaco 9, Tabs at 3 spaces ************************************************************** ** SCSI-Managing Control Panel by ** Horst Pralow ** All rights reserved ** © Horst Pralow, 1991/up to now ** ** Changes History ************************************************************** No changes history recorded for versions older than 1.0d8 Version 1.0d8: Improved Error checking before trying to read data from a specific device. Astonishingly this sometimes causes the cdev to hang where it worked correctly on previous Versions. Version 1.0d9: 03/08/92 Addressed the beforementioned problem. Added a call to TestUnitReady before trying to operate on the device. Added support for old style (Inside Macintosh Vol. IV) Partitioning scheme. Not tested yet. The cdev did definitly crash on such drives in the earlier versions. Version 1.0d9.1: 03/10/92 Fixed problems with InsideMac vol IV style drives and revised partition info display. Now the partitioning scheme is shown when no single partition is selected. Version 1.0b1: 03/11/92 Fixed problems during the mount phase and reload of a driver. Included a call to FlushDataCache on 68030 machines before calling the new driver. Version 1.0.b2: 04/17/92 Added color support and changed the way the cdev does its initial display when first opened. Also removed the Option-Key from "Start/Stop-Unit" It now determines for the current id whether to Start or Stop the unit by calling TestUnitReady. Version 1.0b3: 04/26/92 Improved handling of update events in main window (No more flicker when hitting an item) Added updatehandler for child dialogs. Ballon help implemented. Version 1.0b4: 05/01/92 Improved user feedback on keyboard commands Added display of watch cursor for mount and start/stop Refined ModalFilter function for child dialogs: Return and Enter-keys are handled correct now. Multiple-driver bug addressed (testing will show if it was fixed). We now look for the ddType field to determine which driver to load. A corresponding bug in the CompareDrivers routine fixed. We now compare the Driver-name in the DCTLEntry^ with the driver loaded from disk and not with the one already im memory (this was of course pure bullshit). Version 1.0b5: 05/07/92 Mount support for old style drivers added which expect a pointer to the device partition map (type TS) in A0 on entry. Minor update bugs in main window fixed. Version 1.0b6: 05/13/92 Added special case treatment for drivers like FWB's HDT bunch of drivers. These ones use a special stage one loader and so driver compare never worked. Added expand- dable resource of type 'Scas' which contains substrings to search for in case CompareDrivers failes. If this substring is found in the driver or loader (we can not know about it) we'll treat CompareDrivers successfull. We now pass a pointer to disk block 1 for all drivers in A0 on calling the driver. HDT drivers rely on this feature where as most drivers do not. Version 1.0b7: 05/25/92 Added 'Options'-Button and a pre-version of the Options- Dialog. Only options-group "Stop-Options" works. Additional checking before accessing any device included. Stop-Options: On stop-unit we try to umnount all parti- tions by default now. If there are partitions busy then the ones umnounted will be remounted invisible for the user. Version 1.0b7.1: 05/31/92 Added balloon help for Options Button and Options dialog Version 1.0b7.2: Stop-Unit with Auto-Unmount lost watch-cursor cause PBUnMount implicitely calls InitCursor. We now restore the wristwatch until the drive has stopped and returned completion. Version 1.0b7.3: Implemented mount option with Built-in SCSI driver. Modified driver to be compatible with old-style IM IV partitioning scheme. Version 1.0b7.4: Internal version / new bugs introduced Version 1.0b7.5: Pop-Up menus in main window introduced. They show all available HFS partitions on the selected drive. INIT for mounting Hotkey (without mounting function yet) Introduced. Communaction cdev/INIT via PPCToolbox in System 7. Minor bug fixes Version 1.0b7.6: PopUpMenus now selectivly mounts the choosen volume. Version 1.0b8: First version with working mount hotkey. Thus big leap in version numbering. Mount hotkey is not yet configurable. Default: Command-Control-M. Version 1.0b8.1: Changed the popup menu display to include the total volume size of each partition displayed. Also show a check mark for the partitons already mounted Version 1.0b8.2: Improved appearance on color monitors by using the users highlight color in certain places. Bug fixes according to crashes when using the popup menu feature. Version 1.0b8.3: 01/27/93 Adding self test for Code-Movement. Remove Control-Track- king when deactivated. Version 1.0b8.4: 03/10/93 SCSI-code totally rewritten to be more compact and also to address potential problems with Virtual Memory. If Virtual Memory is running we now hold all our buffers we use during a SCSI transaction. Finally: a method to configure the hotkey Animated cursor changed. Now Implemted via a VBL-Task Removed the option "Unmount only if Option-down". From my opinion this option doesn't make sense. Known glitches: Typos in help balloons. Also: Help balloon not changed to reflect the changes in the Options dialog. Some glitches in the display of the hotkey if it is one of the keypad keys. Still a need to propose the changes to the SCSI code to the INIT. Potential problems with the driver: Alhough the driver runs stable and reliable on virtually all tested drives there still are theoretical problems with virtual memory. I have tested the driver with VM and never run into problems but there might be difficulties using VM with MTE's built in driver and VM because the driver is not reentrant on an partion level. This will be fixed in the next ß-release. Another problem arises from the two SCSI busses of the Quadra 9xx family of Macs. I haven't found any documenta- tion yet how to distinguish between the two SCSI busses. so MTE still goes the old fashioned way to assume a single bus with ID in the range from 0 to 7. Version 1.0b8.5: 03/14/93 Cosmetic fixes: Item size for displaying the hotkey chan- ged. Fixed the display of certain hotkeys (all F-keys on the USS Saratoga*, and others) which were not dis- played correctly. Fixed "Can't find partioning infor- mation" bug. Enhanced busy cursor routine now using an 'acur' resource. (Modify it if you like. MTE will use up to 10 amination frames defined in the 'acur' resource. Changed the file’s creator type to 'MΔEª'. Linked the INIT with a new version of Paul Mercer’s ShowInit to support color icon display at boot time. No major changes. The whole weekend was dedicated to my Porsche 914 03/19/93 Updatet the INIT to share the same SCSI code with the cdev (eg. code being certified VM aware) Additional checking if A/UX is running. The cdev has always checked if _SCSIDispatch is implemented. On A/UX its documented not to be implemented. This worked fine in A/UX 2.0 and 2.01 (where SCSIProbe blindly crashed, haha) With A/UX 3.0 things have changed however: The trap is implemented!! But all it does is calling the SystemError function. So MTE crashed too. We now do an additional Gestalt and refuse to run if gestaltAUXVersion is any- thing but 0. Apple should have left things the way they used to be. There is no sense in implementing a trap just to call the bomb dialog (because a program would get there anyway if it tried to use that trap), but it makes checking and writing clean code a lot more difficult. 03/29/93 Fixed some problems with the INIT which were introduced due to bug fixes to the THINK RSRCRuntime.lib I made according to a TechNote from Symantec. These led to crashes when the cdev tried to read the Init’s current settings. Now I have fixed Symantec’s fixes. These were nothing but another bug! 04/15/93 New Version of the SCSI-driver included (V. 3.7) The driver now is completely reentrant thus addressing theoriticly possible problems with Virtual Memory. (On various test none of these problems arose, but were possible according to Apple information.) These problems are fixed now. For the Tech folks: The problem was when the driver served an I/O request from the File Manager a page-fault could occur and then the driver would have to be able to service the VM-Manager request to page logical memory without forgetting (i.e. overwriting) the context of the interrupted File I/O. For my opinion this hardly can happen in real life. There is one strong argument for my point of view (apart from practical experience): The SCSI Manager itself is single threaded in its current imple- mentation. So when the situation statet above (Page fault during SCSI I/O would appear, the Mac will bomb anyway, no matter how VM arawre and reentrant the driver is). Apple itself is aware of this problem and has the the Device Manager to hold all memory incorporated with the ongoing I/O transaction in advance. If the SCSI-Manager were reentrant (thus fully VM-Compatible) they would not have to do this. But for now we can be shure of: When the Mac crashes with VM its not my driver's fault, it might be the ROM. Changed behaviour of PopUp Menu in main dialog. With a removable media device in case there is no media in the device MtE displays a Menu telling "Insert Media" instead of just doing nothing Fixed a few typos in this file. Know outstanding bugs: Sometimes the display in the main window gets confused and draws some weird boxes instead of legable characters if MTE encounters an error in SCSI communication. Some strange things using Norton Utilities with MTE open were reported to me, but I could not reproduce the bug. It seems rather unlikely to me having any problems with a program like Norton. There is only one trap-patch in the code and this only gets activated when starting a SCSI driver just read from disk to be more friendly to old drivers who would cause a system crash under System 7 if they where loaded without the patch. These drivers still can be used with Mt. Everything (e.g. one of your ancient initialized cartridges you haven't used for years). After installing the driver the patch is removed immediately and no other program gets CPU time between installation and removal of the patch. Balloon help isn't fixed yet and still is somewhat sluggish. Last note: I encountered a bug on a IIci with an 14" Color Monitor set to 8 bit color depth: On this machine opening Mt. Everything and clicking on any SCSI item in the main window dropped me into MacsBug with an Illeagl Instruction error. Having the window deactiveted an then activated again, everything worked fine. I was unable to reproduce this error on any other machine. If you see this misbehaviour too, please give me a note, and I'll investigate on it. Possible future improvements: Additional Stop Unit option to eject the drives media on removable media devices. Support for large block size foramatted disk i.e. > 512 bytes per block. Fix for the Check Disk Bug discussed recently in MacWeek. Implemtation of the INIT/cdev communication via Gestalt to be able to change INIT-settings without restart when running System 6 (6.0.4 at least) Final note: Please don't use the version of MTE found an the Sytem 7 Utilities CD-ROM published recently. That version is fairly outdated. Version 1.0b8.6: 06/09/93 Fixed a bug with the Cmd-Key Popup menu that caused the menu "Insert media" to appear with non-removable drives when the drive in question was stopped. The menu no longer appears with fixed disks. 06/14/93 Fixed a nonproper if-then-else expression which caused the hotkey to fail when "Use Driver found on disk" was selected and the hotkey had to load that driver because it wasn't in RAM yet. Found a bug in the formac Manager software (Versions 4.0 through 4.2 were tested) which caused my built-in driver to use wrong partition boundaries on formac-ini- tialized cartridge drives. The result is the "Disk is unreadable" dialog in the finder. This is because formac stores certain private data in the DataStart and DataCnt field of a partition map entry thus confusing my driver. The cdev however uses a more elaborated code to check for the location of partitions and displays correct information. This problem will probably be addressed in the next release. (Since we name Mt. Everything and not Mt. A Lot of Things.) Version 1.0b8.7: 25.6.1993 Altered the pop-Up menu code so that the additional Item "All Volumes" only appears if multiple partitions con- taining HFS volumes are offline. Previously the item was shown on each drive that contained multiple partitions of whom at least one was offline. 06/27/1993 Enhanced version of the SCSI-Driver. The new driver does some additional checking when identifying HFS-partitions. This should fix problems with formac's Version 4 remov- able drivers but it is a general reliability enhancement. Version 1.0b8.8: 07/21/1993 The last enhancement of the driver introduced a severe bug that caused the driver to report a wrong partition size and to put false entries into the drive queue. This led to problems with Disk First Aid (it reported that the volume could not be read or that the volume is not an HFS Disk). Worse this bug could destroy all data on a drive if the user were to select the Erase Disk command in the Finder on one of the Volumes. SO DO NOT USE VERSION 1.0b8.7 Fixed a bug that caused MtE to stop a drive even if one or more volumes on that drive were busy when the user has selected the Auto-Unmount and Close Driver options 07/23/1993 Updated Balloon help 07/24/1993 Changed the code in the info Dialog which displays the partitins on a drive. It used to display partitions in the order in which they are defined in the partition map map partition. But since the appearence of the list imposeses to show things in the order they are located on the driver I changed the code to do so. The info Dialog now gives more verbose info when an A/UX partition is selected. (This information is not always correct but uses the structures found on the disk as documented by Apple. The A/UX dp program shows the same incorrect information.) Version 1.0b9: 09/12/93 Revamped the INIT code. Made it to squeeze 300 bytes of code. INITS should be as small as possible. Also changed the code for the options dialog slightly to make it more compact and more readable (at least for me). 11/01/93 Finally a limited support for removable media eject: The Stop-Unit button now tells the drive to allow media removal before stopping the drive. Version 1.0b10: 12/12/93 Fixed a bug in the popup - menu code that assigned a command key equivalent to a partitions menu if the volume name happened to contain a "/" character. Changed the appearence of the title Picture and its hilighting when clicked. Version 1.0b11: 12/26/93 Revamped the appearence of the header Picture to fully support multiple Monitors with different PixelSizes to always display the best looking Pict even when the window scans multiple monitors. New feature: The Startup Disk is underlined with a Gray line. And you can set a different startup disk with the Boot Disk button. This roughly is the same as in the Startup Disk control panel. Version 1.0b12: 02/06/94 Mt. Everything finally runs on A/UX 3.0. None of our snizzy features is available right now but at least the main window lists the connected drives Fixed a bug in the INIT code that caused the crossover icon we display in case of loading errors to be placed on the next free icon position rather then on top of ours Known bug: When selecting a specific partition from a device's pop-up menu then Mte will mount all Partitions of the device instead of only the requested one. This happens when Mte must install the driver for the device in order to mount any partition. Once the driver is installed the popup menu works as expected. 02/27/94 Modified A/UX calls to use undocumented ioctl calls which work more reliable than the former attempt through /dev/gd 03/01/94 The pop-up menu now mounts a specific partition from a drive even if the driver for that drive was not loaded when the user made the menu selection. (This fixes bug noted 6.2.94). 03/10/94 The pop-up menu no longer changes the current scsi id if the mouse is released outside the menu. Revamped some segmentation issues of the INIT. 03/13/94 Unix support is now fully implemented. At least the Mac portion of it. There is not (yet) a way to mount/umount Unix-File Systems. But since A/UX offers commands to do this it's not such a miss. There also is no Start/Stop when running UNIX cuz the Kernel wouldn't like it! The cdev has now been made multi-segment with all the A/UX stuff in its own seg since I was rapidly approaching the 32 K limit the linker insists on. Brand new About-Box! (this is a major feature in some commercial packages!) Version 1.0f1 03/29/94 This is final candidate 1. Fixed several System 6 related bugs: The modifiers field in the event record passed to the cdev seems to contain garbage. So I test for modifier keys myself when running system 6. Fixed the dialog autopositioning code to produce (nearly) the same results when running sys 6 as when running sys 7. i.e. the modal dialogs now appear at the same location where they would appear in system 7. Improved color support. 04/13/94 Fixed some mistakes when displaying partition info for an A/UX partition (slice number and date). Note that still the dates may be incorrect. That is because the impplementation of A/UX does not behave as documented! Verson 1.0f2 05/01/94 Fixed some typos and added items to balloon help. Fixed a bug in the SCSI-Transaction code which could cause an address-error on 68000 macs under certain condi- tions (Yep, twas a word operation on an odd address. MC68000s dont like this). Discovered problems with my old Mac SE which I sold years ago and which I had the pleasure to get my hands on recently. That machine is powered by an 16 MHz 020 card and Mt. Everything behaves bit strange. It did not crash but showed wrong information regarding wether a drive is ready or not. An earlier version which did much less error-checking than we do now seemed to work better. Since I had no opportunity to test I only guess my Test- Unit-Ready somehow fails on that machine. Is it the SE- ROM? Is it the accelerator card or its software? How does one tell without Macsbug installed? 05/04/94 Added limited suport for DAT-Drives. Now Mt. Everyting is able to eject DAT cartridges by mouse click. I did it cuz I needed it. If anybody can urge strongly for a specific feature, I'll think about it. (No promises) During the meanwhilst [see Monty Python "The Life of Brian (soundtrack)" for further reference] while doing a total rebuild on my Porsche 914 (everything I didn't fix last winter) I started work on Mt. Every- thing 2. It's a total rewrite in C++ (for the sake of PowerPC portabilty, not that I like C++ that much. Quote: C++ isn't a standard. Its a moving target). Features will include: Resizable window, support for SCSI-Manager 4.3 (i.e. multiple SCSI busses), save- and printable device information with more info to be available. It's new, it's OOPs and it will eat lots more of your RAM and it will be System 7 only. 5/8/94 Fixed a timout problem when ejecting a DAT tape. When the drive didin't manage to eject the tape within the sup- plied time the SCSI Manager would lock and the Mac would hang when doing a subsequent disk i/o. Version 1.0 (final to be released) 9/9/94 After some month of delay due to the hot summer and my vacancies I discovered a bug while writing the documen- tation for MTE. It’s the Reload command (option Mount). This one did not check wether a drive was busy but rather did close an installed driver from a busy disk and in- stalled a new copy of the driver. It would do this even on the disk containing the boot volume or worse the Virtual Memory backing file! (shudder) Appearently none of the users tried to do this since I never received a bug report. But finally this one is fixed. 10/09/94 After just another month being very busy in my job where I get paid for being busy I added some integrety checks to the code. Just in case… The documentation matures. 10/12/94 Added a keyboars shortcut for Reload. Its Option-Shif-R 11/17/94 Removed shortcut for Reload since its not compatible on international keyboards and one should use it rarely. Now the modal dialogs handle update events for background windows (at least for the cdev window) User documentation is ready. 02/01/95 Version 1.0.1 Fixed a noncrucial bug in the SCSI-Code and made some optimizations. Fixed some glitches in ballon help. 5/10/95 Fixed some ugly typos and bad stylistics in the docu- mentation. Version 1.0.2 08/04/95 Revamped mechanism to call popup-menu to fix some spurious crashes encountered while using the feature. No one did report on this bug but on my own machine problems occured frequently. 09/18/95 In V. 1.0 I introduced some code to check for modifica- tions made on certain parts of Mt.E. This was due to the fact that in a german Mac magazine I once foumd a Control Panel which was a stolen early apha version of Mt.E that had been modified in a way to show a different "About" picture and a different author. I wasn ’t very pleased to see someone steal my work so I dicided to fix this once and for all. I introduced self checking code that verifies Mt. Everthing is still the same as it left my desk and would destroy the running copy if t found modifications that would not please me, the author. This has benefits and drawbacks: One of the benefits is a built-in virus check. If some virulent code tries to hook into Mt.E it will notice and warn you. The drawback of course is there are situations where other code modifies my code: OmegaSANE is an example. This resulted sometime to Mt.E reporting a modification to one of its resources and destroying itself where everything was perfectectly legal. The dog simply was a little too sharp. I revised this sceme to be more user friendly. Mt.E will no longer destroy itself but still will give you a warning alert in case it finds modifications to its own resources. It is now OmegaSANE compatible. Hackers still will have a time! Version 1.0.3 10/16/1995 Ported code to compile with Universal Interfaces 2.1. Here we are back again on compatibility road! 10/18/1995 Fixed some System 6 related problems in screen update and revamped parts of the overall appearance. Streamlined the code for Partition info dialog but its still a mess. This one really needs a rewrite! Disabled "Stop-Unit" command for A/UX. Was always inten- ded to be so but in fact wasn't. 10/22/1995 Partition info now is more verbose with A/UX partitions In case multiple drivers are found on a disk, the default one now gets marked. SCSI Mgr 4.3 drivers now will be displayed as such. Fixed a bug in my UNIX SCSI-code which luckily was im- possible to ever show on any users machine but mine! 10/29/95 Added styled display of SCSI standard conformance in main window. Made this feature switchable. Therefor redesigned the options dialog. 11/10/95 through 11/19/95 Added an discussion about SCSI Manager 4.3 to documen- tation and overhauled the rest of it. ************************************************************** *USS Saratoga was reported to be the code name of the Apple Extended Keyboard, just in case you didn't know ************************************************************** ***END OF FILE***