[9front] drawterm lib configuration

[9front] drawterm lib configuration

[Romano]

I use drawterm from different OSes to connect to my
9front systems and noticed that there's a generic
pattern that I usually take in modifying my lib/profile
for different clients that I'm drawterm'ing from.
I also noticed that the newuser(1) man page's
presentation of what is generated for lib/profile was
out-dated and not all architectures had underlying 'bin'
directories created for the user. So I figured I'd take
a stab at updating newuser(1) and its man page to have
newuser also create a lib/drawterm directory, with
lib/drawterm/default being the corresponding profile for
when someone drawterm's in to the system. Here's a
link to my attempt in case anyone finds it useful:

http://only9fans.com/unobe/patches/4ac3a0224ed9d54818f858fba69c8e94f38f2c12/9front/2235c398fa9e7b48e0c84cda05c6994a14736e55.patch/raw

Re: [9front] drawterm lib configuration

[Jacob Moody]

On 8/27/24 13:40, Romano wrote:
> I use drawterm from different OSes to connect to my
> 9front systems and noticed that there's a generic
> pattern that I usually take in modifying my lib/profile
> for different clients that I'm drawterm'ing from.
> I also noticed that the newuser(1) man page's
> presentation of what is generated for lib/profile was
> out-dated and not all architectures had underlying 'bin'
> directories created for the user. So I figured I'd take
> a stab at updating newuser(1) and its man page to have
> newuser also create a lib/drawterm directory, with
> lib/drawterm/default being the corresponding profile for
> when someone drawterm's in to the system. Here's a
> link to my attempt in case anyone finds it useful:
> 
> http://only9fans.com/unobe/patches/4ac3a0224ed9d54818f858fba69c8e94f38f2c12/9front/2235c398fa9e7b48e0c84cda05c6994a14736e55.patch/raw

I know this isn't a request to merge this in to 9front, but I still wanted to comment on some things I noticed in your patch.
The whole lib/drawterm/* thing doesn't seem like it would be too useful to me personally in general.

diff c32dabd4853888f62f09a6d3f8e0deed4077b6a7 2235c398fa9e7b48e0c84cda05c6994a14736e55
--- a/sys/lib/newuser
+++ b/sys/lib/newuser
@@ -1,5 +1,4 @@
 #!/bin/rc
-
 user=`{cat /dev/user}
 home=/usr/$user
 if(test -f $home/lib/profile){

This seems like accidental noise that got in to your patch.

@@ -8,10 +7,10 @@
 }
 cd $home
 x='$'
-mkdir bin bin/rc bin/mips bin/386 bin/amd64 bin/power bin/arm bin/arm64
-mkdir lib tmp
+mkdir bin/^('' rc spim arm arm64 amd64 386 power power64 mips)
+mkdir lib lib/drawterm tmp
 chmod +t tmp
-bind -qc /n/other/usr/$user/tmp $home/tmp
+if(test -d /n/other/usr/$user/tmp) bind -qc /n/other/usr/$user/tmp $home/tmp
 bind -c $home/tmp /tmp
 mail -c
 auth/cron -c
@@ -22,38 +21,41 @@

You can use mkdir -p instead to clean up these calls, and remove this bin/('') thing.
It'll read more naturally.

@@ -22,38 +21,41 @@
 font=/lib/font/bit/vga/unicode.font
 switch($x^service){
 case terminal
-	webcookies
-	webfs
-	plumber
 	echo -n accelerated > '#m/mousectl'
 	echo -n 'res 3' > '#m/mousectl'
 	prompt=('term% ' '	')
 	fn term%{ $x^* }
+	webcookies
+	webfs
+	plumber
 	rio

What does moving this around achieve?

 case cpu
-	bind /mnt/term/dev/cons /dev/cons
-	bind -q /mnt/term/dev/consctl /dev/consctl
-	>[2] /dev/null {
-		cp /dev/sysname /mnt/term/dev/label
-		if(wsys=`{cat /mnt/term/env/wsys} && ~ $x^#wsys 1) {
-			wsys=/mnt/term^$x^wsys
-		}
-		if not {
-			wsys=()
-		}
-	}
-	bind -a /mnt/term/dev /dev
+	# if rcpu or drawterm:
+	if(test -d /mnt/term/dev){
+		bind /mnt/term/dev/cons /dev/cons
+		bind -q /mnt/term/dev/consctl /dev/consctl
+		>[2] /dev/null {
+			cp /dev/sysname /mnt/term/dev/label
+			if(wsys=`{cat /mnt/term/env/wsys} && ~ $x^#wsys 1)
+				wsys=/mnt/term^$x^wsys
+			if not
+				wsys=()
+		}
+		bind -a /mnt/term/dev /dev
+	}
 	prompt=('cpu% ' '	')
 	fn cpu%{ $x^* }
-	if(! test -e /mnt/term/dev/wsys){
-		# call from drawterm
-		if(test -e /mnt/term/dev/secstore){
-			auth/factotum -n
-			read -m /mnt/term/dev/secstore >/mnt/factotum/ctl
-			echo >/mnt/term/dev/secstore
-		}
-		if not
-			auth/factotum

The reason this check is written this way is because drawterm will not have a /mnt/term/dev on windows.
So your first check here will fail with a windows client.


- moody

Re: [9front] drawterm lib configuration

[Romano]

[-- Attachment #1: Type: text/plain, Size: 4676 bytes --]

Thanks for the feedback, moody. My inline explanations below.

On Tue, Aug 27, 2024, at 3:05 PM, Jacob Moody wrote:
> On 8/27/24 13:40, Romano wrote:
>> I use drawterm from different OSes to connect to my
>> 9front systems and noticed that there's a generic
>> pattern that I usually take in modifying my lib/profile
>> for different clients that I'm drawterm'ing from.
>> I also noticed that the newuser(1) man page's
>> presentation of what is generated for lib/profile was
>> out-dated and not all architectures had underlying 'bin'
>> directories created for the user. So I figured I'd take
>> a stab at updating newuser(1) and its man page to have
>> newuser also create a lib/drawterm directory, with
>> lib/drawterm/default being the corresponding profile for
>> when someone drawterm's in to the system. Here's a
>> link to my attempt in case anyone finds it useful:
>> 
>> http://only9fans.com/unobe/patches/4ac3a0224ed9d54818f858fba69c8e94f38f2c12/9front/2235c398fa9e7b48e0c84cda05c6994a14736e55.patch/raw
>
> I know this isn't a request to merge this in to 9front, but I still 
> wanted to comment on some things I noticed in your patch.
> The whole lib/drawterm/* thing doesn't seem like it would be too useful 
> to me personally in general.

It helps me setup my drawterm defaults based on what client I am dialing in from (e.g. work laptop, my wife's laptop, my macbook). I anticipate having more but also I don't assume it's useful for everyone.

> diff c32dabd4853888f62f09a6d3f8e0deed4077b6a7 
> 2235c398fa9e7b48e0c84cda05c6994a14736e55
> --- a/sys/lib/newuser
> +++ b/sys/lib/newuser
> @@ -1,5 +1,4 @@
>  #!/bin/rc
> -
>  user=`{cat /dev/user}
>  home=/usr/$user
>  if(test -f $home/lib/profile){
>
> This seems like accidental noise that got in to your patch.

I didn't see a point in the blank line.

>
> @@ -8,10 +7,10 @@
>  }
>  cd $home
>  x='$'
> -mkdir bin bin/rc bin/mips bin/386 bin/amd64 bin/power bin/arm bin/arm64
> -mkdir lib tmp
> +mkdir bin/^('' rc spim arm arm64 amd64 386 power power64 mips)
> +mkdir lib lib/drawterm tmp
>  chmod +t tmp
> -bind -qc /n/other/usr/$user/tmp $home/tmp
> +if(test -d /n/other/usr/$user/tmp) bind -qc /n/other/usr/$user/tmp $home/tmp
>  bind -c $home/tmp /tmp
>  mail -c
>  auth/cron -c
> @@ -22,38 +21,41 @@
>
> You can use mkdir -p instead to clean up these calls, and remove this 
> bin/('') thing.
> It'll read more naturally.

Thanks! I had considered that but didn't know if I wanted to force directory creation and that's why it wasn't done before. So I opted for at least a list expansion.

> @@ -22,38 +21,41 @@
>  font=/lib/font/bit/vga/unicode.font
>  switch($x^service){
>  case terminal
> -	webcookies
> -	webfs
> -	plumber
>  	echo -n accelerated > '#m/mousectl'
>  	echo -n 'res 3' > '#m/mousectl'
>  	prompt=('term% ' '	')
>  	fn term%{ $x^* }
> +	webcookies
> +	webfs
> +	plumber
>  	rio
>
> What does moving this around achieve?

I didn't think it was foolish to have consistency with the other case (cpu) to have those run at the end. I didn't see the point of having them before those other commands.

>  case cpu
> -	bind /mnt/term/dev/cons /dev/cons
> -	bind -q /mnt/term/dev/consctl /dev/consctl
> -	>[2] /dev/null {
> -		cp /dev/sysname /mnt/term/dev/label
> -		if(wsys=`{cat /mnt/term/env/wsys} && ~ $x^#wsys 1) {
> -			wsys=/mnt/term^$x^wsys
> -		}
> -		if not {
> -			wsys=()
> -		}
> -	}
> -	bind -a /mnt/term/dev /dev
> +	# if rcpu or drawterm:
> +	if(test -d /mnt/term/dev){
> +		bind /mnt/term/dev/cons /dev/cons
> +		bind -q /mnt/term/dev/consctl /dev/consctl
> +		>[2] /dev/null {
> +			cp /dev/sysname /mnt/term/dev/label
> +			if(wsys=`{cat /mnt/term/env/wsys} && ~ $x^#wsys 1)
> +				wsys=/mnt/term^$x^wsys
> +			if not
> +				wsys=()
> +		}
> +		bind -a /mnt/term/dev /dev
> +	}
>  	prompt=('cpu% ' '	')
>  	fn cpu%{ $x^* }
> -	if(! test -e /mnt/term/dev/wsys){
> -		# call from drawterm
> -		if(test -e /mnt/term/dev/secstore){
> -			auth/factotum -n
> -			read -m /mnt/term/dev/secstore >/mnt/factotum/ctl
> -			echo >/mnt/term/dev/secstore
> -		}
> -		if not
> -			auth/factotum
>
> The reason this check is written this way is because drawterm will not 
> have a /mnt/term/dev on windows.
> So your first check here will fail with a windows client.

I was surpised by this. It just so happens that the Windows recovery USB stick (which cost $20) for an ideapad I got for $5 came in today. So I installed Windows 10 Home and downloaded drawterm.exe; I see a /mnt/term/dev directory (see the attached screenshot), and as I drawterm'd in, my patch works as I expected. Is it perhaps different for different versions of Windows? Does your drawterm.exe not provide a dev dir?

[-- Attachment #2: IMG_5538.jpeg --]
[-- Type: image/jpeg, Size: 1544724 bytes --]

Re: [9front] drawterm lib configuration

[Romano]

[-- Attachment #1: Type: text/plain, Size: 5016 bytes --]

I do not know why the image I attached is corrupted. Trying with a smaller image size.

On Wed, Aug 28, 2024, at 1:15 AM, Romano wrote:
> Thanks for the feedback, moody. My inline explanations below.
>
> On Tue, Aug 27, 2024, at 3:05 PM, Jacob Moody wrote:
>> On 8/27/24 13:40, Romano wrote:
>>> I use drawterm from different OSes to connect to my
>>> 9front systems and noticed that there's a generic
>>> pattern that I usually take in modifying my lib/profile
>>> for different clients that I'm drawterm'ing from.
>>> I also noticed that the newuser(1) man page's
>>> presentation of what is generated for lib/profile was
>>> out-dated and not all architectures had underlying 'bin'
>>> directories created for the user. So I figured I'd take
>>> a stab at updating newuser(1) and its man page to have
>>> newuser also create a lib/drawterm directory, with
>>> lib/drawterm/default being the corresponding profile for
>>> when someone drawterm's in to the system. Here's a
>>> link to my attempt in case anyone finds it useful:
>>> 
>>> http://only9fans.com/unobe/patches/4ac3a0224ed9d54818f858fba69c8e94f38f2c12/9front/2235c398fa9e7b48e0c84cda05c6994a14736e55.patch/raw
>>
>> I know this isn't a request to merge this in to 9front, but I still 
>> wanted to comment on some things I noticed in your patch.
>> The whole lib/drawterm/* thing doesn't seem like it would be too useful 
>> to me personally in general.
>
> It helps me setup my drawterm defaults based on what client I am 
> dialing in from (e.g. work laptop, my wife's laptop, my macbook). I 
> anticipate having more but also I don't assume it's useful for everyone.
>
>> diff c32dabd4853888f62f09a6d3f8e0deed4077b6a7 
>> 2235c398fa9e7b48e0c84cda05c6994a14736e55
>> --- a/sys/lib/newuser
>> +++ b/sys/lib/newuser
>> @@ -1,5 +1,4 @@
>>  #!/bin/rc
>> -
>>  user=`{cat /dev/user}
>>  home=/usr/$user
>>  if(test -f $home/lib/profile){
>>
>> This seems like accidental noise that got in to your patch.
>
> I didn't see a point in the blank line.
>
>>
>> @@ -8,10 +7,10 @@
>>  }
>>  cd $home
>>  x='$'
>> -mkdir bin bin/rc bin/mips bin/386 bin/amd64 bin/power bin/arm bin/arm64
>> -mkdir lib tmp
>> +mkdir bin/^('' rc spim arm arm64 amd64 386 power power64 mips)
>> +mkdir lib lib/drawterm tmp
>>  chmod +t tmp
>> -bind -qc /n/other/usr/$user/tmp $home/tmp
>> +if(test -d /n/other/usr/$user/tmp) bind -qc /n/other/usr/$user/tmp $home/tmp
>>  bind -c $home/tmp /tmp
>>  mail -c
>>  auth/cron -c
>> @@ -22,38 +21,41 @@
>>
>> You can use mkdir -p instead to clean up these calls, and remove this 
>> bin/('') thing.
>> It'll read more naturally.
>
> Thanks! I had considered that but didn't know if I wanted to force 
> directory creation and that's why it wasn't done before. So I opted for 
> at least a list expansion.
>
>> @@ -22,38 +21,41 @@
>>  font=/lib/font/bit/vga/unicode.font
>>  switch($x^service){
>>  case terminal
>> -	webcookies
>> -	webfs
>> -	plumber
>>  	echo -n accelerated > '#m/mousectl'
>>  	echo -n 'res 3' > '#m/mousectl'
>>  	prompt=('term% ' '	')
>>  	fn term%{ $x^* }
>> +	webcookies
>> +	webfs
>> +	plumber
>>  	rio
>>
>> What does moving this around achieve?
>
> I didn't think it was foolish to have consistency with the other case 
> (cpu) to have those run at the end. I didn't see the point of having 
> them before those other commands.
>
>>  case cpu
>> -	bind /mnt/term/dev/cons /dev/cons
>> -	bind -q /mnt/term/dev/consctl /dev/consctl
>> -	>[2] /dev/null {
>> -		cp /dev/sysname /mnt/term/dev/label
>> -		if(wsys=`{cat /mnt/term/env/wsys} && ~ $x^#wsys 1) {
>> -			wsys=/mnt/term^$x^wsys
>> -		}
>> -		if not {
>> -			wsys=()
>> -		}
>> -	}
>> -	bind -a /mnt/term/dev /dev
>> +	# if rcpu or drawterm:
>> +	if(test -d /mnt/term/dev){
>> +		bind /mnt/term/dev/cons /dev/cons
>> +		bind -q /mnt/term/dev/consctl /dev/consctl
>> +		>[2] /dev/null {
>> +			cp /dev/sysname /mnt/term/dev/label
>> +			if(wsys=`{cat /mnt/term/env/wsys} && ~ $x^#wsys 1)
>> +				wsys=/mnt/term^$x^wsys
>> +			if not
>> +				wsys=()
>> +		}
>> +		bind -a /mnt/term/dev /dev
>> +	}
>>  	prompt=('cpu% ' '	')
>>  	fn cpu%{ $x^* }
>> -	if(! test -e /mnt/term/dev/wsys){
>> -		# call from drawterm
>> -		if(test -e /mnt/term/dev/secstore){
>> -			auth/factotum -n
>> -			read -m /mnt/term/dev/secstore >/mnt/factotum/ctl
>> -			echo >/mnt/term/dev/secstore
>> -		}
>> -		if not
>> -			auth/factotum
>>
>> The reason this check is written this way is because drawterm will not 
>> have a /mnt/term/dev on windows.
>> So your first check here will fail with a windows client.
>
> I was surpised by this. It just so happens that the Windows recovery 
> USB stick (which cost $20) for an ideapad I got for $5 came in today. 
> So I installed Windows 10 Home and downloaded drawterm.exe; I see a 
> /mnt/term/dev directory (see the attached screenshot), and as I 
> drawterm'd in, my patch works as I expected. Is it perhaps different 
> for different versions of Windows? Does your drawterm.exe not provide a 
> dev dir?
> Attachments:
> * IMG_5538.jpeg

[-- Attachment #2: File1.png --]
[-- Type: image/png, Size: 526149 bytes --]

Re: [9front] drawterm lib configuration

[Jacob Moody]

On 8/28/24 03:30, Romano wrote:
> I do not know why the image I attached is corrupted. Trying with a smaller image size.
> 
> On Wed, Aug 28, 2024, at 1:15 AM, Romano wrote:
>> Thanks for the feedback, moody. My inline explanations below.
>>
>> On Tue, Aug 27, 2024, at 3:05 PM, Jacob Moody wrote:
>>> On 8/27/24 13:40, Romano wrote:
>>>> I use drawterm from different OSes to connect to my
>>>> 9front systems and noticed that there's a generic
>>>> pattern that I usually take in modifying my lib/profile
>>>> for different clients that I'm drawterm'ing from.
>>>> I also noticed that the newuser(1) man page's
>>>> presentation of what is generated for lib/profile was
>>>> out-dated and not all architectures had underlying 'bin'
>>>> directories created for the user. So I figured I'd take
>>>> a stab at updating newuser(1) and its man page to have
>>>> newuser also create a lib/drawterm directory, with
>>>> lib/drawterm/default being the corresponding profile for
>>>> when someone drawterm's in to the system. Here's a
>>>> link to my attempt in case anyone finds it useful:
>>>>
>>>> http://only9fans.com/unobe/patches/4ac3a0224ed9d54818f858fba69c8e94f38f2c12/9front/2235c398fa9e7b48e0c84cda05c6994a14736e55.patch/raw
>>>
>>> I know this isn't a request to merge this in to 9front, but I still 
>>> wanted to comment on some things I noticed in your patch.
>>> The whole lib/drawterm/* thing doesn't seem like it would be too useful 
>>> to me personally in general.
>>
>> It helps me setup my drawterm defaults based on what client I am 
>> dialing in from (e.g. work laptop, my wife's laptop, my macbook). I 
>> anticipate having more but also I don't assume it's useful for everyone.

What defaults are you changing here?

>>
>>> diff c32dabd4853888f62f09a6d3f8e0deed4077b6a7 
>>> 2235c398fa9e7b48e0c84cda05c6994a14736e55
>>> --- a/sys/lib/newuser
>>> +++ b/sys/lib/newuser
>>> @@ -1,5 +1,4 @@
>>>  #!/bin/rc
>>> -
>>>  user=`{cat /dev/user}
>>>  home=/usr/$user
>>>  if(test -f $home/lib/profile){
>>>
>>> This seems like accidental noise that got in to your patch.
>>
>> I didn't see a point in the blank line.

Most of our scripts have this blank line, it's style.

>>
>>>
>>> @@ -8,10 +7,10 @@
>>>  }
>>>  cd $home
>>>  x='$'
>>> -mkdir bin bin/rc bin/mips bin/386 bin/amd64 bin/power bin/arm bin/arm64
>>> -mkdir lib tmp
>>> +mkdir bin/^('' rc spim arm arm64 amd64 386 power power64 mips)
>>> +mkdir lib lib/drawterm tmp
>>>  chmod +t tmp
>>> -bind -qc /n/other/usr/$user/tmp $home/tmp
>>> +if(test -d /n/other/usr/$user/tmp) bind -qc /n/other/usr/$user/tmp $home/tmp
>>>  bind -c $home/tmp /tmp
>>>  mail -c
>>>  auth/cron -c
>>> @@ -22,38 +21,41 @@
>>>
>>> You can use mkdir -p instead to clean up these calls, and remove this 
>>> bin/('') thing.
>>> It'll read more naturally.
>>
>> Thanks! I had considered that but didn't know if I wanted to force 
>> directory creation and that's why it wasn't done before. So I opted for 
>> at least a list expansion.

What do you mean "forcing directory creation" you're calling mkdir,
you're creating directories. The two calls with -p or not with your
ordering are equivalent.

>>
>>> @@ -22,38 +21,41 @@
>>>  font=/lib/font/bit/vga/unicode.font
>>>  switch($x^service){
>>>  case terminal
>>> -	webcookies
>>> -	webfs
>>> -	plumber
>>>  	echo -n accelerated > '#m/mousectl'
>>>  	echo -n 'res 3' > '#m/mousectl'
>>>  	prompt=('term% ' '	')
>>>  	fn term%{ $x^* }
>>> +	webcookies
>>> +	webfs
>>> +	plumber
>>>  	rio
>>>
>>> What does moving this around achieve?
>>
>> I didn't think it was foolish to have consistency with the other case 
>> (cpu) to have those run at the end. I didn't see the point of having 
>> them before those other commands.

The point is to avoid churn, but I'm bikeshedding.

>>
>>>  case cpu
>>> -	bind /mnt/term/dev/cons /dev/cons
>>> -	bind -q /mnt/term/dev/consctl /dev/consctl
>>> -	>[2] /dev/null {
>>> -		cp /dev/sysname /mnt/term/dev/label
>>> -		if(wsys=`{cat /mnt/term/env/wsys} && ~ $x^#wsys 1) {
>>> -			wsys=/mnt/term^$x^wsys
>>> -		}
>>> -		if not {
>>> -			wsys=()
>>> -		}
>>> -	}
>>> -	bind -a /mnt/term/dev /dev
>>> +	# if rcpu or drawterm:
>>> +	if(test -d /mnt/term/dev){
>>> +		bind /mnt/term/dev/cons /dev/cons
>>> +		bind -q /mnt/term/dev/consctl /dev/consctl
>>> +		>[2] /dev/null {
>>> +			cp /dev/sysname /mnt/term/dev/label
>>> +			if(wsys=`{cat /mnt/term/env/wsys} && ~ $x^#wsys 1)
>>> +				wsys=/mnt/term^$x^wsys
>>> +			if not
>>> +				wsys=()
>>> +		}
>>> +		bind -a /mnt/term/dev /dev
>>> +	}
>>>  	prompt=('cpu% ' '	')
>>>  	fn cpu%{ $x^* }
>>> -	if(! test -e /mnt/term/dev/wsys){
>>> -		# call from drawterm
>>> -		if(test -e /mnt/term/dev/secstore){
>>> -			auth/factotum -n
>>> -			read -m /mnt/term/dev/secstore >/mnt/factotum/ctl
>>> -			echo >/mnt/term/dev/secstore
>>> -		}
>>> -		if not
>>> -			auth/factotum
>>>
>>> The reason this check is written this way is because drawterm will not 
>>> have a /mnt/term/dev on windows.
>>> So your first check here will fail with a windows client.
>>
>> I was surpised by this. It just so happens that the Windows recovery 
>> USB stick (which cost $20) for an ideapad I got for $5 came in today. 
>> So I installed Windows 10 Home and downloaded drawterm.exe; I see a 
>> /mnt/term/dev directory (see the attached screenshot), and as I 
>> drawterm'd in, my patch works as I expected. Is it perhaps different 
>> for different versions of Windows? Does your drawterm.exe not provide a 
>> dev dir?

Right, drawterm does add it's own /dev/ to the root before exposing the
namesapce to the remote system so you'll see drawterm's /dev.
My recollection of the ordering was wrong and in retrospect doesn't make sense.

That first check is just checking to see if we should make use of the /mnt/term/dev anyway,
so it seems fine to have that be the check there.

Re: [9front] drawterm lib configuration

[Romano]

On Wed Aug 28 07:31:20 -0700 2024, moody@posixcafe.org wrote:
> What defaults are you changing here?

Setting up different plumbing rules, different fonts
(remember that bad patch for rio border size?), certain
mounts. For example, when I drawterm from a mac I use
osascript to open a new tab for web links, and open any docx
files on the Mac. I also create a 9pfs mount (as explained on
the wiki) to browse files using the Finder, if I prefer.

> Most of our scripts have this blank line, it's style.

Thanks for pointing that out. I see 77/135 of the scripts in
/bin/rc have a second blank line, so 57% overall:
cpu% { for (f in `{walk -f /rc/bin}) { echo `{sed 2q $f | tail -1} } } | sort | uniq -c | sort -r | sed 4q
  77 
  25 rfork e
   3 rfork en
   2 if(~ $#* 3){
cpu% ls /rc/bin | wc -l
    135

This isn't comprehensive by any means, but just a directory I knew
had a lot of system rc scripts. For comparison, /bin/git directory (like /sys/lib/newuser) isn't included in the numbers above:
cpu% { for (f in `{walk -f /bin/git}) { echo `{sed 2q $f | tail -1} } } | sort | uniq -c | sort -r | sed 3q
   8 rfork ne
   5 rfork en
   3 

In your experience, is calling rfork on the second line an exception to
that style? If so, that would make about 78% of the scripts in /rc/bin
conform to the style. But removing the '| sed 4q' at the end, you'd also
see quite a few first lines being explanatory comments of what the script
does, and about a dozen that just call exec.

I haven't checked out a Plan 9's 4th edition or previous edition to
compare to see if it has been newer style, or just always been a
hodge-podge.

Regardless, it's good for me to be aware of that: thank you.

> What do you mean "forcing directory creation" you're calling mkdir,
> you're creating directories. The two calls with -p or not with your
> ordering are equivalent.

Forcing intermediate directory creation, so really the 'bin' dir. This
just reflected how the long-form was before my patch: it didn't use -p.
I didn't know if there was a good reason for that, and thought it might
have been for some esoteric but valid reason related to creating bin
first, before its subdirectories. So I decided to write it without -p.

> >>> @@ -22,38 +21,41 @@
> >>>  font=/lib/font/bit/vga/unicode.font
> >>>  switch($x^service){
> >>>  case terminal
> >>> -	webcookies
> >>> -	webfs
> >>> -	plumber
> >>>  	echo -n accelerated > '#m/mousectl'
> >>>  	echo -n 'res 3' > '#m/mousectl'
> >>>  	prompt=('term% ' '	')
> >>>  	fn term%{ $x^* }
> >>> +	webcookies
> >>> +	webfs
> >>> +	plumber
> >>>  	rio
> >>>
> >>> What does moving this around achieve?
> >>
> >> I didn't think it was foolish to have consistency with the other case 
> >> (cpu) to have those run at the end. I didn't see the point of having 
> >> them before those other commands.
> 
> The point is to avoid churn, but I'm bikeshedding.

I'm not sure it's bikeshedding here. My perspective is that a script should
not do the same thing in a different order unless called for. I was trying
to determine why the first three calls were separated from the last, when
that isn't the case for cpu. I assume I will read this script again, and so
considered it a benefit for the human reader to consolidate the calls as is
done in the cpu case.

> Right, drawterm does add it's own /dev/ to the root before exposing the
> namesapce to the remote system so you'll see drawterm's /dev.
> My recollection of the ordering was wrong and in retrospect doesn't make sense.
> 
> That first check is just checking to see if we should make use of the /mnt/term/dev anyway,
> so it seems fine to have that be the check there.

*Whew*, well I'm glad that was one less thing I had possibly fumbled!

Thanks again for the feedback, moody.