[Info-vax] DCL, was: Re: Microkernel

[David Froble]

Simon Clubley wrote:
> On 2012-08-12, FrankS <sapienza at noesys.com> wrote:
>> At the risk of taking this thread even further off into tangents unknown,
>> could someone explain the BENEFIT of being able to recall commands from past
>> sessions?  It seems like a lot of work to implement (BASH or DCL) and I can't
>> remember any time I needed to recall a command from a prior session that I
>> couldn't retype from scratch.  I mean, really, how complicated are the command
>> lines you're building?
>>
> 
> It's not only the length of the commands, but also having quick and easy
> ways of recalling them and constructing them. In bash, you can type Ctrl-R
> followed by a few characters and keep hitting Ctrl-R until the matching
> command is found. Tab completion makes it very easy to construct long
> command lines as well.
> 
> This means that instead of having lots of little files floating around
> with one line commands in them, it's _much_ easier to use the command
> history file to hold the commands.
> 
> Generally, when I am working on something, I work on it for more than one
> day. I also have a series of commands I use on a regular basis. In either
> case, I can just type a few characters per command the next day and have
> the command available once again.
> 
> Using a command history file is self maintaining as well. When I have not
> used the commands for a while, they will automatically be removed from
> the history file without me having to delete anything.
> 
> You asked for some example commands. I cannot discuss anything I am doing
> at work, but below are some examples from home. I am currently doing some
> work on a USB device stack of mine and it's very easy to find the related
> sets of commands and PDF documents when I pick it up the next evening.
> [I have a large (but organised) reference library and with tab completion
> it's as quick to find a document from the command line as it is to use
> a GUI browser, especially when you can recall the command and use it to
> find related documents or use locate or find on the reference library
> tree.]
> 
> stty -F /dev/ttyS3 raw -echo time 60 min 0 -crtscts -hupcl
> 
> xpdf /reflib/arm_arch_docs/cpu_docs/arm7dtmi/atmel/doc6175_AT91SAM7S_DS.pdf &
> 
> The following command is one command; I've broken it into several lines.
> There are also variants for burning to flash and for different images.
> There are also related commands for running ddd (and specifying the
> required gdb backend) on the command line. It's _very_ easy to recall
> one of these variants the next day with a few characters.
> 
> FNAME=/sd/arm-test/usb-dev-h256/usb-dev-h256_sram.bin
> 	/projs2/embedded/exe-openocd-0.4.0/bin/openocd
> 	-f /sd/simuclib/scripts/arm/sam7s256/openocd-pp-ram-run.cmd
> 
>> Generally speaking, for my purposes, command recall more than a dozen or so
>> commands back in the current session has been unnecessary.  And even then I'm
>> just being too lazy to type it in again.
> 
> With a permanent command history, easy recall of commands ($ recall/search
> does _NOT_ qualify as easy recall) and tab completion of filenames, it's
> something I really miss even on repeated short commands when it's not
> available. If it's anything more complicated than viewing a file in the
> current directory, it's quicker to get it from the command history under
> bash.
> 
> Simon.
> 

Well, frankly, I'm with Frank on this one ....

(Ok, stop with the boos and hisses)

I feel that command recall is very useful, but, only for the last 20-40 
commands.  If you have some complex commands that you perform often, why not 
have a command file you can invoke?  Yes, that is about what a permanent recall 
buffer might be considered.  But if the command is complex enough, the command 
file is, to me anyway, easier to maintain.

I understand that VMS is used in many ways.  Just when I think I've seen all of 
them, I get my nose rubbed in another.  So I'm open to some usage somewhere, but 
in general, I see it (permanent command buffer) as something that gets used 
because some feel it is useful, but I just don't see it.

In the traditional VT terminal session, you're communicating with a computer. 
What I'm saying today is what I have to say today.  If talking to a person, I 
don't keep saying "what I said yesterday, what I said 2 months ago, what I said 
3 weeks ago, ....".  I just say what I have to say right now.  I just feel that 
at some point it can be easier to just "say" what you have to say than finding a 
command from a hugh recall buffer.  I just don't comprehend a thousand or more 
commands that I'm going to retrieve and modify and use over and over.  If I have 
a procedure that I want to use periodically, I build a command file and perhaps 
have it re-queue itself in a batch queue.

I also sometimes find the editing of a command more difficult than just typing 
the desired command.

About the most use I make of command recall is when I'm working on a program and 
not doing so well.  I might use the up-arrow key to keep repeating maybe 4-5 
commands, EDT, COMPILE, LINK, PURGE, RUN.  Even then, it can be a close call 
between recalling commands and just re-typing them.