Please make it stop

> On Feb 15, 2019, at 10:56 PM, Brian White <[email protected]> wrote:
> 
> By tightly coupled, I mean that Telcom has inside  knowledge of the rest of 
> the system.
> 
> The tpdd does not.
> 
> (nor the dvi, and I'd put money on the chipmunk too)
> 
> The only reason you can produce examples of other software and systems doing 
> things like that, is because it's something that used to be very common, and 
> even today every new programmer takes some time to learn better, and some 
> never do. But by now it is recognized as an antipattern. After 
> Joe-self-taught-teenager-programmer-from-the-70s-80s writes a bunch of such 
> stuff, it's all fine at first, and only later he discovers that he has 
> written an unmaintainable and unportable brick. Customer wants the tiniest 
> change and it's practically impossible, because the code is riddled with 
> assumptions that should never have been made.
> 
> 
> John did raise some points that no tpdd emulator to date has emulated the 
> drive *really* though, else they would use disk images instead of files. And 
> the knitting machines don't use files either.
> 
> I personally don't think that's a strong enough argument, but it is at least 
> an argument. I think that just means a tpdd emulator is a filesystem instead 
> of a disk. OK, so I would not tolerate a filesystem that modified the files 
> either.
> 
> 
> As to Mikes point about causing harm:
> 
> Preserving data is never causing harm.
> 
> If there is harmful data, then the harm was caused by the thing that created 
> that harmful data.
> 
> I see that as "two wrongs don't make a right".
> 
> If a text editor added eof, that was a bad text editor, and it's more helpful 
> to the user to have the bad data and thus indirectly the cause of the bad 
> data, be exposed and corrected, rather than silently allowed to persist and 
> even propagate over time. So that you go on thinking you actually have good 
> data when you don't and it just fails tomorrow in any other context.
> 
> -- 
> bkw
> 
> 
> On Fri, Feb 15, 2019, 2:44 PM Gary Weber <[email protected] 
> <mailto:[email protected]> wrote:
>   GW>> Since when is it NOT wrong for TELCOM to disallow recording of <EOF> 
> into a .DO file being downloaded, but it IS wrong for the LaddieAlpha/TS-DOS 
> combination during the same operation?
>   BW>  the fact that telcom is part of a very tightly coupled and managed 
> system. It's not generic software that runs anywhere  
> 
> You have clearly, and strongly, stated the some very sound engineering 
> principles.   Honestly I doubt anyone (or at least many) would question the 
> principles in the universal generic sense.  
> 
> But I would assert LaddieAlpha is not universal or generic.  It has a very 
> defined stated purpose, and has always made specific concessions for Model Ts 
> & WP2s.   Using it outside of this context is at your own peril, so to speak. 
>  And you're bringing up not just edge cases but actually *corner* cases to 
> the usage model.
> 
> Secondly, TELCOM doesn't know where data is coming from via the COM: port, so 
> I question the claim that it is part of a very tightly coupled and managed 
> system in the context of how it downloads data.  Rather, it filters the data 
> to be appropriate to the machine.  The LaddieAlpha/TS-DOS coupling is no more 
> nefarious than this, and to knowingly permit that combination to crash the 
> machine, while TELCOM protects the machine, would not only be silly, it would 
> be irresponsible.  
> 
> Unless the user directs it to do so of course, whereby my 
> --allow_file_system_corruption switch would suffice.  ;-) 
> 
> 
> On Fri, Feb 15, 2019 at 11:31 AM Brian White <[email protected] 
> <mailto:[email protected]>> wrote:
> All else being equal, it IS wrong for telcom not to record whatever bytes are 
> sent.
> 
> It's excuse is, at the time such ideas hadn't been recognized and discarded 
> yet, and the fact that telcom is part of a very tightly coupled and managed 
> system. It's not generic software that runs anywhere, it's part of a rom, and 
> so it's less of a crime to be tightly coupled with the other parts of that 
> rom, like making presumptions about what kind of data may exist in a file 
> with a certain kind of name. 
> 
> The tpdd makes no such presumptions.
> 
> Telcom also only ever claimed to operate like a user application that deals 
> in text just like a text editor. How the text editor and telcom package up 
> data internally are their own business. They just accept keystrokes and 
> display characters. That is an entirely different domain.
> 
> What you should ask is, when you tell BASIC to read or write from COM:, does 
> BASIC presume to modify the data?
> 
> Similarly would you tolerate a file compressor or other encoder that didn't 
> spit back out exactly the data you put into it?
> 
> Telcom processing the data as though it's text is like an image editor adding 
> metadata on it's own. It's an allowed part of it's job because it's job is to 
> deal in displaying a visual image to you, not to hold that data in trust for 
> something else. IT is the app whose defined role is to manipulate and process 
> the data.
> 
> But then when you take that image data the the image editor made, and put it 
> into a zip file, and save it to a thumb drive, and then read it back from a 
> thumb drive, it's entirely wrong for any of those things to output one single 
> bit different from what they were given.
> 
> On Fri, Feb 15, 2019, 5:19 AM Gary Weber <[email protected] 
> <mailto:[email protected]> wrote:
> Since when is it NOT wrong for TELCOM to disallow recording of <EOF> into a 
> .DO file being downloaded, but it IS wrong for the LaddieAlpha/TS-DOS 
> combination during the same operation?
> 
> On Fri, Feb 15, 2019 at 2:29 AM Brian White <[email protected] 
> <mailto:[email protected]>> wrote:
> Since when does a disk drive decide what is and is not supposed to be treated 
> as text vs binary data? The application author who decides which flags to use 
> with open() does that, or the ftp client user does that by using the bin 
> option *which exists*. The the drive subsystem never ever decides that.
> 
> No one answered the question about the sewing machines which also used these 
> drives.
> 
> "AFAIK "good programming practices and principles" are suggestions and 
> recommendations, not laws."
> Indeed.
> 1) I never said anything else.
> 2) Why is a "good practice" good in the first place? For a reason, not 
> because god said so or something. Something is good or bad because of the 
> harm it lessens or increases.
> 
> 
> 
> 
> -- 
> Gary Weber
> [email protected] <mailto:[email protected]>
> 
> 
> -- 
> Gary Weber
> [email protected] <mailto:[email protected]>

Reply via email to