Re: Ada Tutor Web Site Shutting Down

["Randy Brukardt" <randy@rrsoftware.com>]

"Rugxulo" <rugxulo@gmail.com> wrote in message 
news:c31b2969-0e79-44c6-b351-77e1adeb3406@z13g2000prk.googlegroups.com...
>On May 11, 7:45 pm, "Randy Brukardt" <ra...@rrsoftware.com> wrote:
>> "Rugxulo" <rugx...@gmail.com> wrote in message
>>
>> news:291504a4-ec55-45f1-bf7f-13078bf71c3e@m10g2000yqd.googlegroups.com...
>>
>> >On May 5, 4:12 pm, "Randy Brukardt" <ra...@rrsoftware.com> wrote:
>>
>> >> Better, dunno. Janus/Ada 95 (32-bit only) would be updated if anyone 
>> >> wanted
> >> it, but no one has asked. (It's still in our pricelists.)
>>Updated how?
>>
>> Just recompiled and tested. It might need some runtime updates, but 
>> probably
>> not a lot.
>
>What for? Improvements or just to remove the bitrot?

To bring it to the same level of support as the rest of our compilers, of 
course. If no one wants new features, we'd be happy to send them the old 
one...

>> All of our Ada 95/2005/2012 technology is from a common codebase,
>> so building the DOS version just means selecting the right configuration 
>> and
>> recompiling. (And lots of testing and probably a bit of debugging.)
>
>Ah, testing, the biggest time waster. Luckily, Ada has a public test
>suite.

Testing Ada compilers is easy (it's been automated for years). Fixing 
problems uncovered is not so easy.

>> The biggest problem with it is that the DOS extender we used didn't work
>> with NT and afterwards. (It actually works precisely once, then you have 
>> to
>> reboot before it will work again. Not very practical.)
>
>Was it a special homebrew one or one of the popular ones licensed?

Dunno if it was popular, but it was licensed. Something called "Ergo" 
(Pharlap was too expensive to license; the compiler output would work them 
as well, especially good for embedded systems and the like.)

>(What brand name?) Well, either way, I'm not surprised. Half the time
>the extenders were buggy, but the other half was due to XP limitations
>and bugs. (BTW, I assume you meant close/reopen the DOS window, but if
>you meant reboot the computer/OS, that's worse! Ugh.)

No, I meant reboot. The extender does something to the global selector map 
such that subsequent runs fail. It's possible that XP would work better (I 
never tried it on XP), but on NT and Windows 2000 it required a reboot after 
each run before you could run another program with it. It didn't cause any 
other problems (you didn't have to reboot if you didn't want to run some 
other program).

When I reported the problem to the vendor, their response was to go out of 
business. :-) Of course, we didn't have to spend any more money on 
maintenance after that, and since it worked great on Windows 95 and 98, we 
continued to use it well transitioning to all Windows. Once we moved most of 
our desktops to Windows 2000, we moved the development to Windows itself. (I 
think we were pretty much the last DOS customer we had anyway.)

...
...
>> I think 1985 was well before DPMI, and probably was before expanded and
>> extended memory, too. This probably was a basic 640K compiler (I don't 
>> think
>> we even had 640K in our machines at the time - our original IBM PC/XT
>> started with 256K, and I think the Seattle Computer 10 MHZ 8086 [much 
>> faster
>> than the PC] had about that as well).
>
>Right, sorry, I knew that. Though there were 286 extenders by then,
>but the 386 didn't even come out until 1986 or so (and took a while to
>get adopted). Real EMS existed, of course, as did XMS, but it was
>fairly rare. Even in 1994, my new PC only had 4 MB of RAM! My, how
>times have changed.  :-O
>
>But yeah, VCPI (386+) was 1986, DPMI (286 or 386) was 1990. And I hear
>that SCP-8086 could use the full 1 MB (unlike IBM's 640 kb). I assume
>this means yours was a multi-pass compiler (a la Wirth's old M2M /
>Modula-2). But maybe not, perhaps you crammed a lot into "256K" (Ada83
>only probably made it easier).

Our original compilers ran on Z80 CP/M systems -- in as little as 48K of 
memory. Even I can't imagine doing that today, and I know that we did it... 
Obviously, we put as much on disk as we could, but since the disks were 
floppies, that wasn't a perfect solution either.

The Seattle S-100 system had two 1.2M floppies. They were more practical 
than early hard disks, because you had unlimited storage. I usually kept one 
disk for the OS, compiler, basic tools, and commonly shared files, and the 
other disk had the source code and object files for whatever program or part 
of a program I was working on. Lots of disk swapping, but otherwise worked 
great.

>> >Unfortunately, Windows has gone downhill (in DOS support) since XP,
>>
>> You could have just left out the parenthetical remark, and you still 
>> would
>> have been right. ;-)
>
>More or less, yes. I know MS thinks XP is old and wants to deprecate
>it, but it seemed fine to me. But I'm sure some fringes had various
>complaints about it that only Vista or 7 solved. I'm not Windows-y
>enough to know, honestly. (But I do think they charged too much for
>Vista -> 7 ["minor"] upgrade.)

I don't want to have to learn a new interface every couple of months, which 
is the Google approach to software. (We can move buttons around anytime we 
want.) Sadly, people like Firefox are trying to copy that brain-damaged 
approach. (Firefox 4 seems to have moved things mostly for the sake of being 
different. It took me a couple of days to figure out where "Organize 
Bookmarks" and "Reload page" had gone. Grumble.

...
>> That's one of the reasons I'm not moving most of my work to Windows 7
>> anytime soon. I still use a number of DOS programs regularly, including a
>> circa 1986 programming editor.
>
>I'm surprised, honestly, most people spit on the idea of (still) using
>DOS software. But not me!  ;-)   Windows 7 32-bit probably (somewhat)
>works for DOS, as I know Vista 32-bit barely did (some bugs, DPMI
>registry hack needed, no full screen gfx).
>
>Though I wonder at using such an old editor, but I know people like
>what they like, so I can't complain. (Text editors are a dime a dozen
>but none do everything, not even Emacs or VIM. Check out 
>http://www.texteditors.org
>sometime for a laugh.)

Reasons: familar (to me at least) keyboard interface, no mouse fru-fru, 
optional rectangle marking (great for editing text in columns), find/replace 
in any selection (including rectangles), list view for find (shows all 
matches at once, great for finding stuff when you aren't sure what you're 
looking for), convinient macro facilities (by recording).

Disadvantages: DOS :-), limited window size (80x25 isn't always big enough), 
limited fonts (whetever you can adjust the console window to show), little 
integration with Ada, bugs cause crashes (but of course I know what not to 
do, so it those only happen when I'm careless).

I'd build a replacement using Claw and Ada if I had time, but I wonder if 
the result would really work as well.

                          Randy.