ORIGINALLY PUBLISHED IN LIMA NEWSLETTER DECEMBER 1993 LETTER from AUSTRALIA - No. 6 Sep / 93 ----------------------------------------- A week or so ago we were down in Sydney for the weekend visiting William. Nothing for it but we had to go out with him to see Jurassic Park which had opened in Australia just a few days before. This was his second time at this movie, the first being a few days before on its second day with the whole office where he works - yes, all 24 of them, with more than just passing interest as graphics programmers and animators. This leads as I found to comments on things like imperfections in matting in of the dinosaurs. All mightily impressed though. This is not a movie review here, but it is worth noting that unlike most movie and TV efforts showing computers which range from fantasy to excruciating, this one looked for real in scenes where people were using computers even to regular Silicon Graphics users like Will. This extended to running the computers in the film with modified video boards at 48 Hz frame rate and synchronized to the film camera (film uses 24 frames/sec) to eliminate the moving black bands or heavy flicker normally seen on TV films of actual computer screens. Your editor Charlie should take well to a movie in which almost the first word uttered is "palaeobotanist". Last weekend was not a happy one at Funnelweb Farm. We had not seen our resident mother possum for some days, and on Saturday morning Val heard a noise at the front door, and there was mother with a now fairly large baby on her back climbing painfully up the front screen door to a temporarily open space a couple of meters deep between the floors. She had been savaged by a dog but the little one was unharmed. We managed to snaffle junior, now old enough to be semi-independent - you wear garden gloves while doing so and pop the little critter in a hessian potato sack. Then you can let it peek out while holding the bag, which I ended up holding this time. It took a lot of coaxing to get mother possum to drag herself back out of the hole. When she did, Val, with gloves and sack of course, was able to bring her down and hold her quietly. Brush-tail possums usually bite and scratch and can do a lot of damage while being able to wriggle out of the tightest grip, but this one was so far gone that she could only lie quietly in Val's arms and take some water from a spoon, even some banana being too difficult, now she could see that junior was safe. The lady from the Native Animal Trust came around and took mother possum and son off to the vet. She was very relieved that we had captured the possums first. The sad but expected later news was that mother possum was too far gone to survive, but at least the little one will get a chance. I am deeply angry at irresponsible dog owners. It is supposed to be an offense to allow a dog to harm protected native animals, but if you do not catch it in the act then there is little that can be proved. Some have wondered why we have this fixation on funnelweb spiders around here. Just recently there was a repeat showing of an excellent TV natural history special on spiders, mostly Australian. I expect that this would have or will be shown in the US of A as National Geographic was one of the partners in making it, so if you have any curiosity in this area do not miss it when it comes to a TV set near you. The program is called "Webs of Intrigue" and is presented by Densey Cline. Some things seem to be the same no matter what type of computer you have - suppliers who are incompetent or shonky in varying degrees. It is not that many years ago that even IBM threatened to leave Australia if computers were brought under consumer law that insisted that computers compute as claimed. Car warranties just depress me! I bought at considerable expense a Canadian product for my home 486 PC, an ATI Graphics Ultra Plus video card. Though it took a second try to get one that would pass its own self-diagnostics (the supplier all the while denying the first one was faulty), the hardware generally has a good reputation, though murmurs are now surfacing about that too. What ATI definitely appear to be unwilling to do or incompetent at is software. Now video handling with DOS on PCs is a disgusting mess, with separate drivers seemingly needed for every major DOS program. Maybe ATI has done those adequately, but when you get to a more abstracted level the picture is not so good. and the OS/2 support is abysmally bad. Despite ATI advertising OS/2 support, what has emerged has been incomplete, very late, and buggy. I am not yet game to install the latest offerings. So like most OS/2 users with these cards I scale back to the 8514/A simulation with the IBM driver, while all that very expensive video co-processor remains effectively unused. I shall certainly avoid buying any ATI product again. Over recent years I have bought quite a number of PCs at work, and it seems that the process of getting them up and running on all fronts is becoming steadily more traumatic. If you don't test them thoroughly then you may find yourself in post-warranty trouble. Murphy guarantees it. Some have a theory that you rarely if ever find both hardware and software competence in the same company. That's how Will got into the Amiga/SCSI business, by working as the programmer with a hardware designer who needed software expertise to make a real team. And while on the theme of shonky suppliers, Will has several kilobucks of his part time earnings while a student tied up with DMI, a San Diego area video board firm which never delivered the product and is dragging forever on refunding the $$. While we are on this subject I should mention that I have a unknown, but perhaps very unusual distinction in the TI scene. Years ago I actually got my money back from 99er Magazine! Maybe I just got in early enough or made the right kind of noises. We had relied on borrowed copies in the early days, and the editorial content though never very impressive did seem to be improving in the run-up to the 99/8, which I was waiting for impatiently. So I took out a subscription, and then about one or two much delayed issues later it looked like losing the plot, so I jumped in and cancelled. Which brings me to the strange case of OPA. There do seem to be some serious unresolved matters here. My friends in the Sydney UG ordered a whole pile of TIM/SOBs, paid for and never delivered - this has badly damaged the TI scene in Australia. It seems though that Gary Bowser just refuses to communicate about the matter, not even to deny any liability. I have tried recently on behalf of Sydney also with no response - I know the e-mail did not bounce. Yet the strange thing is that OPA still appears on the networks and at fairs and offers products, even, I am told including the TIM on occasion. There seems to be an attitude expressed in that strange letter from OPA that appeared in MicroPendium that as money paid for orders a couple of years ago went to pay living expenses at the time, then customers should not expect to receive product. Customers might expect otherwise, and even Ontario law cannot be so wide open as to countenance such goings on. From what I have seen of the one TIM/SOB that made it to the Hunter, it was indeed a good product, and the Vn 8.1x HRD ROS as supplied by Bud Mills has been a real contribution, despite its limitations. Trouble is that nothing else OPA has produced in a long while now seems of any real significance. On the TI front at Funnelweb Farm, there is both good news and bad. The bad is not intrinsically so, just that I have not found much time or energy for programming or much else on the TI since releasing the Vn 5 Editor. The good news is that after Geoff Trott's latest ministrations, the HRD-3000 at long last seems to be behaving itself. I may even work up to getting some more 128Kx8 SRAMs, if local suppliers have heard of them yet, to bring it up to a more useful size. Am I game yet to retire a 192Kb HRD to duty as a memory expansion test bed? Maybe it is time to look at patching the ROS to handle DSQD equivalent. With the all-seeing benefit of hindsight, it now seems a great pity that the HRD design did not originally include both ROM and RAM in the initial DSR space. It needs both, ROM at >4000 including power-up check routines for ruggedness and SRAM to carry the configurable information that has to be accessible to a standard DSRlink search and for more working room if needed. The fragility of the DSR header has always plagued HRD users, and the attempt at an EPROM sold in the past was not really successful. I have had a browse through the Asgard AMS memory expansion user guide on disk, but any opinions of this memory expansion scheme formed and expressed here are from that and net-news items only, and not from actual experience. The memory expansion scheme itself appears quite livable with, paging in 4K blocks from a large pool into the usual TI low and high memory spaces. It will not reproduce the full function of TI's own 128K expansion, but seeing as these were never released and there were so few anyway it hardly matters, and it is otherwise more extensible by addition of more memory. It does reside at a fixed CRU base, but this is up high at >1E00 and so avoids the low range crush (the TI lived at >1200). I do question the lack of a DSR. This is promoted as a virtue, but that seems like trying to make a feature out of a deficiency, a traditional sport of marketeers. No doubt it saves board area and is cheaper to make. Why would the user want to have a DSR even if it cost more? One reason would be to provide self-test routines accessible from console Basic without requiring even the minimal 32K expansion to be operational (this is one saving grace of the Myarc 512K Ramdisk card). Other desiderata might be an identification call, and a library of routines either callable in place or downloadable preferably as position independent code. No doubt more can be added to this list. I would agree that the full DSRlink mechanism for memory bank selection as in RAMBO on HRD cards is more overhead than desirable, but existence of a DSRlinked control does not imply that more direct access to the low level banking mechanisms is not possible. The access just needs to be clearly and permanently defined, and preferably simple and elegant. As remarked in a previous Letter the real problem with banked memory expansion is the decision overhead in handling large data blocks. The programmer with more or less trouble can structure code to work in bank sized blocks with little run-time overhead. These things are not trivial either - look at how Microsoft with all their billions of dollars and armies of programmers have not been able to make 16-bit Windoze work without too frequent crashes. Then again lots of things are nicer on the TI which is why some still like playing with it. I am not quite so keen on some of the software decisions made in the AMS system (mostly those to do with PAD usage and which may have been forced by the absence of working RAM in the DSR space). A great deal of work has gone into it, and it may be when the RAG Macro-assembler and Linker finally come into their own, in these expanses of paged memory. I have always thought a macro-assembler was gross overkill and even counterproductive in the standard 32K TI-99 environment, but it and the enhanced Linker may well be essential tools for AMS. I do not care at all for the way areas of PAD have been appropriated, in particular the violation of the E/A manual specification (p406) of >83D0-2 as DSR search pointer storage. The TI system specifications are very clear on just what PAD memory a DSR is allowed to trash (some cards like the Myarc FDC use more but restore it on exit). A hardware device that uses the CRU banking scheme for DSRs and claims for itself usage of memory not allocated to such devices is making a claim to ownership of the whole system instead of being one amongst equals. A memory expansion is special, but not all that special. Putting aside all the rival commercial claims which were coming hot and heavy a while back, it is clear that the TI-99 has long needed CPU memory expansion. Even when the original 99/4 was conceived, a plain 64Kb address space was clearly not enough for its designers, but the general CPU memory expansion never made it out the door before Black Friday. This has ever since cramped the style of programmers and limited the scope of applications. We are never going to have the luxury of wide open linear address spaces on the 16-bit 9900 series CPUs no matter how extra memory is paged in, but the 99/4a system can support a substantial amount of extra CPU memory. The 9995 CPU in the Myarc Geneve can realistically support even more just because it is faster, despite its on-chip memory block not being in a very suitable place for 99/4 emulation. Just as an aside I learned recently that the 9995s used in the 99/8 were specials with on-chip memory disabled. Back to the main train of thought -- Myarc for their own reasons never made flexible memory expansion available on the 99/4a, limiting it to the generally difficult to use all-at-once scheme of the 512Kb RamDisk card, even though the technical means now being used in the AMS card were available before then. Realistically at this stage in the life of the 99/4a the market can support at most one such device, and the AMS card appears to have center stage, and I wish them well. Note that I say "appears" as I have seen no external evidence of it yet as a real commercial product. Since writing the last Letter I have been able to access Delphi and pop up on the TI section every so often under the name GLOBAL01. This may be at very odd times because I telnet via the Internet into Delphi from my office machine, at lunch time or after work Australian time, which tends to make it very late evening or early AM in USA. The connection from Australia is usually pretty hesitant - I don't know what Delphi is like on its home ground - but an extra layer of packetized communication across the Pacific can only make it deteriorate, and I often have a long delay before keystrokes register on my screen. I think it has improved a little since the link out of Australia moved from satellite to cable. Sometimes initial access times out waiting for the password. Each time I check any mail and read the Forum entries. So if you have any queries feel free to make them by this channel, and I will answer them there as best I can. If any warnings or easy bug-fixes come to light for FW, I will post them in the Forum. I do not currently have any way to capture whole files, and typing more than short replies directly on Delphi is difficult and inaccurate because of the long key response delays. So any long detailed communications are best sent to my Internet e-mail address if you have that facility. I have already communicated with various people in USA and Europe at their xxx@delphi.com addresses in the last few months, so if you have Internet access my e-mail address is at the end of this letter. Once I forgot to put the .com on delphi at the initial telnet - a local Zeus answered - but thought he was a Vaxstation 3100 and not a greek god. Perhaps I should end with a philosophical puzzle. The other morning I was out for my morning exercise walk in Blackbutt Reserve when I heard a tree fall (real T. Rex sound effect stuff), but even though I walked back down the trail a way, I could not see where it fell. Was it real ? Tony McGovern Funnelweb Farm Sep / 14 / 93 e-mail -- phpam@cc.newcastle.edu.au Delphi -- GLOBAL01 .PL 1