LinkRight 1.1D Patch Users who have had trouble establishing a reliable parallel port connection with LinkRight should get and use this patch. This patch should work much better on those systems. The zip file should unzip to 3 files: LR11D.EXE, LRPAR.SYS and this file, README.TXT. Unzip it to your LINKRT11 directory and then run LR11D.EXE. This is a patch to LinkRight 1.1, so if you have the A, B, or C patches already applied, you'll have to get your original 1.1 diskette, put the LR.EXE and LRCLONER.EXE from your 1.1 diskette into your LINKRT11 directory before running LR11D.EXE. You may have to change the settings of the LinkRight 1.1 object on your desktop to make sure it executes LR.EXE rather than LRBAK.EXE. LinkRight 1.1D will be displayed as the version when you get the first window with the options of Local or Remote so be sure to check for it. Also, LRCLONER displays the version when started, so be sure to check it. When the LRPAR.SYS driver installs, it now displays the version. ****IMPORTANT****** LinkRight 1.1D requires the 1.1D driver. Don't try to use it with the LRPAR from 1.1 or 1.1C, it won't work. Also, a system running 1.1D can only connect to another system that is also running 1.1D. In other words, make sure you update all your executables and all your machines to 1.1D. I assume you can figure out where to place the included LRPAR.SYS. Turbomode (in the Options menu of LinkRight when in Local mode) should be turned off initially. Also, Turbomode is read and set once when you establish a connection. To change Turbomode, change the option, disconnect if you are currently connected, and do a re-connect. If you have no trouble with Turbomode off, you can try Turbomode on and get a speed boost. The LinkRight parallel port driver can now be used in interupt driven mode or polled mode. The default is polled, which matches the default of PRINT01.SYS in OS/2 Warp 3.0 (or so I've been told). Interupt driven mode has some problems. I've had some of my test systems lock up completely when used in Turbomode with interupt mode. The same system does not lock up when Turbomode is off. None of my test systems have any problems when using polled mode. Many gigabytes transferred successfully!! For LRCLONER, you no longer need EAUTIL.EXE and CMD.EXE in the C:\TEMP directory. This should make cloning easier. You still need the temp directoy and LRCLONER.EXE should be started from this directory. Although I said this for 1.1C it turned out not to be true. I really mean it this time!! The complete list of options for LRPAR.SYS are DEVICE=LRPAR.SYS /Q:7 /P /A:3BC /M where: /Q:7 uses IRQ7 or /Q:5 uses IRQ5. These are the only IRQs supported. Interupt driven mode can only be used if PRINT01.SYS or PRINT02.SYS also uses the same interupt. If no /Q option is selected, polled is assumed. /P is for polled. This is a useless option, since it is the default. /A:3BC or /A:378 or /A:278 select the address of the LPT port for LinkRight to use. If this parameter is used, only one address can be specified. Then, when you select Connect from the LinkRight menu, no matter which LPT port you select, either LPT1, LPT2 or LPT3, LinkRight will use the address specified. If this parameter is not included, LinkRight reads the BIOS info at bootup time to get the LPT addresses. /M is to specify MicroChannel machines. For some MicroChannel systems when booted from bootable floppies, LinkRight would not see that it was MicroChannel and would assume an ISA bus. This caused the driver to fail. So if you boot your MicroChannel system from floppies and the LinkRight driver displays a message at bootup time saying an ISA bus was found, use this parameter to fix the problem. The preferred and suggested mode of use is no parameters on the LRPAR.SYS line in CONFIG.SYS. You can add parameters if there are any problems. I'm interested in hearing speed test results on various systems. A quick note would be appreciated. A good test is a 1 megabyte file. Note that a group of files with a total size of 1 megabyte will be significantly slower since there is a lot of overhead for each individual file. For the 1 megabyte file on various systems you should expect these kinds of times. Note that the processor mentioned is for the receiving system and as long as the sending system is close to the speed of the receiving system, the receiving system will be the bottleneck. Please let me know if you get significantly different numbers. 386 SX 16 Mhz 50 seconds 486 SLC 25 Mhz 25 seconds 486 DX 33 Mhz 15 seconds Pentium 60 Mhz 10 seconds These numbers assume 8 bit mode (look at the Remote system when you establish a connection and it will tell you if it is using 8 bit mode or 4 bit mode; LinkRight figures out fastest mode it can use automatically). Expect 30% slower for 4 bit transfers. Also assuming 1024 packet size and Turbomode. Serial mice and slow systems are a problem when using the parallel port. Bump the packet size down to 128 bytes and turn Idle time transfers on and it may improve response with the mouse. PS/2 style mice are not a problem. If you find responsiveness is a problem and you are willing to work with me, I'll create a special driver for your particular systems to improve mouse responsiveness. It may take a few tries to get it right, so patience would be required. Obviously, when I get it working well on your systems, I'll incorporate it into the full LinkRight product. Bump up the Retry Warning Delay option to 9 and you will not get any of those annoying Retry Warnings. LinkRight will try forever if this value is 9. Of course, you won't know if there is a problem either. Major bugs fixed with this version!! There is a bug with the C Set compiler and Toolkit (I think) that caused previous versions of LinkRight to sometimes not copy files or make the receiving file zero bytes. I've worked around this compiler bug and now that I have it fixed for LinkRight, I'll be reporting it to IBM. For the record, IBM may already have a fix for this compiler bug (or I may not be using the API call correctly), so I can't really complain too loudly until I check into this. If you're interested in the details, check the OS2DF1 forum where you'll see a complete report. I used a different API and it seems to work OK now. My thanks to Alan McClean in Vermont and Hans Joachim Lippke in Germany for pointing this bug out to me. Also, my apologies to any users who were bitten by this bug. It was not an easy bug to find (some users may disagree, but they probably just automatically find any bugs) and I had to transfer over 50 Megs at a time to get it to fail. One user complained about updating the Remote system screen just to display a "dot". I didn't see any problem on my systems, and it looked like a minor nit. But during final testing of this version, I grabbed my slowest system, turned the hardware turbo switch off on it, and watched for about a second for it to refresh the screen. Very annoying. There is a workaround available, though. From an OS/2 command prompt, run LRCLONER T LPT1 on the Remote system. You won't lose any functionality and won't be annoyed by the slow screen refresh. For cloning systems, there is an alternative method that should be used in some cases. Boot both systems from bootable floppies and run LRCLONER T LPT1 from one system and LRCLONER S C: LPT1 from the other system. Run LRCLONER with no parameters to check the proper usage. Since both systems are booted from floppies, you don't have to worry about locked files. I've been told you have to do this way if you're using COMM manager or Extended Services. Good luck and happy computing!! Jeff Tremble President Rightware Inc.