20 Jul
2020
20 Jul
'20
5:23 p.m.
As y'all know, I'm a relative latecomer to the world of Unix, but I do
try to figure out how y'all did it back when. So, sometimes, as in this
case, I can figure out how to do something, but I'm curious how it was
done back in the day, moreso than how I can get it done today. I'm
looking at the patching of my shiny new 2.11 BSD pl 431 system running
on my speedy little virtual PDP-11/70, so I grabbed patch 432 (here's a
portion of the patch):
...
To install the update cut where indicated below and save to a file
(/tmp/432) and then:
cd /tmp
sh 432
./432.sh
./432.rm
sh 432.shar
patch -p0 < 432.patch
Watch carefully for any rejected parts of the patch. Failure of a
patch typically means the system was not current on all preceeding
updates _or_ that local modifications have been made.
...
====== cut here
#! /bin/sh
# This is a shell archive, meaning:
# 1. Remove everything above the #! /bin/sh line.
# 2. Save the resulting text in a file.
# 3. Execute the file with /bin/sh (not csh) to create:
# 432.rm
# 432.sh
# 432.shar
# 432.patch
...
# End of shell archive
This seems straightforward. Just follow the directions et voila magic
happens.
My questions for y'all are how would you go about doing this? Use vi to
delete everything through the ==== cut here line? Use some combination
of standard unix tools to chop it up? What was your workflow for
patching up the system using these files?
In my world, if I screw something up, it's 15 seconds to run a restore
script in my simh directory and I can try again, so my level of concern
for a mistake is pretty low. If I was doing this in 1980, on real
hardware, I would have had many concerns, as I'm sure some of y'all can
remember, how did you prepare and protect yourselves so a patch was
successful.
BTW, I thought .shar was an archive format, so when I saw the patch was
a shar file, I was worried it would be in some binary form, lo and
behold, it looks like text to me... not even b64. So much to learn, so
little time.
Thanks,
Will
--
GPG Fingerprint: 68F4 B3BD 1730 555A 4462 7D45 3EAA 5B6D A982 BAAF
20 Jul
20 Jul
5:44 p.m.
On Mon, Jul 20, 2020 at 1:25 PM Will Senn <will.senn(a)gmail.com> wrote:
As y'all know, I'm a relative latecomer to
the world of Unix, but I do try
to figure out how y'all did it back when. So, sometimes, as in this case, I
can figure out how to do something, but I'm curious how it was done back in
the day, moreso than how I can get it done today. I'm looking at the
patching of my shiny new 2.11 BSD pl 431 system running on my speedy little
virtual PDP-11/70, so I grabbed patch 432 (here's a portion of the patch):
...
To install the update cut where indicated below and save to a file
(/tmp/432) and then:
cd /tmp
sh 432
./432.sh
./432.rm
sh 432.shar
patch -p0 < 432.patch
Watch carefully for any rejected parts of the patch. Failure of a
patch typically means the system was not current on all preceeding
updates _or_ that local modifications have been made.
...
====== cut here
#! /bin/sh
# This is a shell archive, meaning:
# 1. Remove everything above the #! /bin/sh line.
# 2. Save the resulting text in a file.
# 3. Execute the file with /bin/sh (not csh) to create:
# 432.rm
# 432.sh
# 432.shar
# 432.patch
...
# End of shell archive
This seems straightforward. Just follow the directions et voila magic
happens.
My questions for y'all are how would you go about doing this? Use vi to
delete everything through the ==== cut here line? Use some combination of
standard unix tools to chop it up? What was your workflow for patching up
the system using these files?
In my world, if I screw something up, it's 15 seconds to run a restore
script in my simh directory and I can try again, so my level of concern for
a mistake is pretty low. If I was doing this in 1980, on real hardware, I
would have had many concerns, as I'm sure some of y'all can remember, how
did you prepare and protect yourselves so a patch was successful.
BTW, I thought .shar was an archive format, so when I saw the patch was a
shar file, I was worried it would be in some binary form, lo and behold, it
looks like text to me... not even b64. So much to learn, so little time.
Thanks,
Will
--
GPG Fingerprint: 68F4 B3BD 1730 555A 4462 7D45 3EAA 5B6D A982 BAAF
5:52 p.m.
On Mon, Jul 20, 2020 at 1:25 PM Will Senn <will.senn(a)gmail.com> wrote:
My questions for y'all are how would you go about
doing this? Use vi to
delete everything through the ==== cut here line?
Yep
In my world, if I screw something up, it's 15
seconds to run a restore
script in my simh directory and I can try again, so my level of concern for
a mistake is pretty low. If I was doing this in 1980, on real hardware, I
would have had many concerns, as I'm sure some of y'all can remember, how
did you prepare and protect yourselves so a patch was successful.
Run an incremental backup and/or copy the files you new you we were messing
with. The good news was that patch makes backups.
BTW, I thought .shar was an archive format, so when I saw the patch was a
shar file,
It was so of. It was a way to send files around that people could easily
execute and you new would work through 7-bit based email which is all the
SMTP guaranteed in the early days. Yeh but .. uucp was 8 yep. But some
of the legs of the USENET were luck to be based on Arpanet site, which
might have had a mailer running BITNET. When shar was created the 'least
needed' style assumptions were used. As it was it was often that people
put tarballs, then compressed them and then uuencoded them inside. Often a
space savings and made it easier -> compressed tar was pretty good, and
even with the 3 8-bit chars as 4 6-bit chars of uuencode it will worked out
well in practice.
5:56 p.m.
On Mon, Jul 20, 2020 at 11:54 AM Clem Cole <clemc(a)ccc.com> wrote:
On Mon, Jul 20, 2020 at 1:25 PM Will Senn <will.senn(a)gmail.com> wrote:
There's various 'unshar' programs, but they are all just restricted
versions of the shell because of the wide diversity of 'shar'
implementation... uuencoded compressed tar balls added another layer to
this mess as well :)
Warner
My questions for y'all are how would you go
about doing this? Use vi to
delete everything through the ==== cut here line?
Yep
In my world, if I screw something up, it's
15 seconds to run a restore
script in my simh directory and I can try again, so my level of concern for
a mistake is pretty low. If I was doing this in 1980, on real hardware, I
would have had many concerns, as I'm sure some of y'all can remember, how
did you prepare and protect yourselves so a patch was successful.
Run an incremental backup and/or copy the files you new you we were
messing with. The good news was that patch makes backups.
BTW, I thought .shar was an archive format, so when I saw the patch was a
shar file,
It was so of. It was a way to send files around that people could easily
execute and you new would work through 7-bit based email which is all the
SMTP guaranteed in the early days. Yeh but .. uucp was 8 yep. But some
of the legs of the USENET were luck to be based on Arpanet site, which
might have had a mailer running BITNET. When shar was created the 'least
needed' style assumptions were used. As it was it was often that people
put tarballs, then compressed them and then uuencoded them inside. Often a
space savings and made it easier -> compressed tar was pretty good, and
even with the 3 8-bit chars as 4 6-bit chars of uuencode it will worked out
well in practice.
6:08 p.m.
shar, by the way, stands for "shell archive." That is, it's an archive
that could be unpacked by feeding the file into sh.
The most complete unshar is probably at
http://sources.vsta.org/comp.sources.unix/volume15/cshar/ It's portable C
(for its time, 20 years ago). Safety, in terms of not trashing an existing
file, was a goal.
On Mon, Jul 20, 2020 at 1:59 PM Warner Losh <imp(a)bsdimp.com> wrote:
On Mon, Jul 20, 2020 at 11:54 AM Clem Cole <clemc(a)ccc.com> wrote:
On Mon, Jul 20, 2020 at 1:25 PM Will Senn <will.senn(a)gmail.com> wrote:
There's various 'unshar' programs, but they are all just restricted
versions of the shell because of the wide diversity of 'shar'
implementation... uuencoded compressed tar balls added another layer to
this mess as well :)
Warner
My questions for y'all are how would you go
about doing this? Use vi to
delete everything through the ==== cut here line?
Yep
In my world, if I screw something up, it's
15 seconds to run a restore
script in my simh directory and I can try again, so my level of concern for
a mistake is pretty low. If I was doing this in 1980, on real hardware, I
would have had many concerns, as I'm sure some of y'all can remember, how
did you prepare and protect yourselves so a patch was successful.
Run an incremental backup and/or copy the files you new you we were
messing with. The good news was that patch makes backups.
BTW, I thought .shar was an archive format, so when I saw the patch was
a shar file,
It was so of. It was a way to send files around that people could easily
execute and you new would work through 7-bit based email which is all the
SMTP guaranteed in the early days. Yeh but .. uucp was 8 yep. But
some of the legs of the USENET were luck to be based on Arpanet site, which
might have had a mailer running BITNET. When shar was created the 'least
needed' style assumptions were used. As it was it was often that people
put tarballs, then compressed them and then uuencoded them inside. Often a
space savings and made it easier -> compressed tar was pretty good, and
even with the 3 8-bit chars as 4 6-bit chars of uuencode it will worked out
well in practice.
6:07 p.m.
On 7/20/20 12:52 PM, Clem Cole wrote:
On Mon, Jul 20, 2020 at 1:25 PM Will Senn <will.senn(a)gmail.com
<mailto:will.senn@gmail.com>> wrote:
My questions for y'all are how would you go about doing this? Use
vi to delete everything through the ==== cut here line?
Yep
Nice, seemed easy enough to me, but I was expecting real Unix folks use
sed | awk | indent type answers.
It was so of. It was a way to send files around that
people could
easily execute and you new would work through 7-bit based email which
is all the SMTP guaranteed in the early days. Yeh but .. uucp was 8
yep. But some of the legs of the USENET were luck to be based on
Arpanet site, which might have had a mailer running BITNET. When shar
was created the 'least needed' style assumptions were used. As it
was it was often that people put tarballs, then compressed them and
then uuencoded them inside. Often a space savings and made it easier
-> compressed tar was pretty good, and even with the 3 8-bit chars as
4 6-bit chars of uuencode it will worked out well in practice.
Hmm... can't wait to run across all of these variants :).
--
GPG Fingerprint: 68F4 B3BD 1730 555A 4462 7D45 3EAA 5B6D A982 BAAF
24 Jul
24 Jul
6:29 p.m.
On Mon, Jul 20, 2020, at 14:07, Will Senn wrote:
On 7/20/20 12:52 PM, Clem Cole wrote:
For whatever it's worth, you can do the exact same thing as vi with sed in this case:
1,/====/d
I wasn't there, so I can't say if this was done in practice or if there was even
enough uniformity in the markers for it to be reliable in theory [perhaps /^#!/,$p ?]
On Mon, Jul 20, 2020 at 1:25 PM Will Senn <will.senn(a)gmail.com> wrote:
Nice, seemed easy enough to me, but I was expecting real Unix folks use
sed | awk | indent type answers. My questions for y'all are how would you go
about doing this? Use vi to delete everything through the ==== cut here line?
Yep
6:46 p.m.
On 7/24/20 1:29 PM, Random832 wrote:
On Mon, Jul 20, 2020, at 14:07, Will Senn wrote:
I'm just getting comfortable using sed, thanks for the simplification.
Will
--
GPG Fingerprint: 68F4 B3BD 1730 555A 4462 7D45 3EAA 5B6D A982 BAAF
On 7/20/20 12:52 PM, Clem Cole wrote:
For whatever it's worth, you can do the
exact same thing as vi with sed in this case: 1,/====/d
I wasn't there, so I can't say if this was done in practice or if there was
even enough uniformity in the markers for it to be reliable in theory [perhaps /^#!/,$p ?]
On Mon, Jul 20, 2020 at 1:25 PM Will Senn <will.senn(a)gmail.com> wrote:
Nice, seemed easy enough to me, but I was expecting real Unix folks use
sed | awk | indent type answers. My questions for y'all are how would you go
about doing this? Use vi to delete everything through the ==== cut here line?
Yep
6:57 p.m.
On Fri, Jul 24, 2020 at 2:47 PM Will Senn <will.senn(a)gmail.com> wrote:
I'm just getting comfortable using sed, thanks
for the simplification.
The issue with sed is that different shar programs used different
cut/here markers.
Using vi was safer and as I said, since you did not do
it everyday, it was no big deal.
Maybe since you have a number of these you want to apply, check with a grep
for the expected marker before you try to apply the sed script and only do
it in a more automated matter catching the vast majority of the markers.]
You asked how we did it. As was said, there were a number of unshar(1)
commands out there, that did the sed for you. Go looking in the archives,
but generally speaking using vi was 'good enough' for day to day work
because you did not run it that often / every day / were >>usually<< not
apply a lot of them in a row.
Clem
7:21 p.m.
On 7/24/20 1:57 PM, Clem Cole wrote:
On Fri, Jul 24, 2020 at 2:47 PM Will Senn <will.senn(a)gmail.com
<mailto:will.senn@gmail.com>> wrote:
Yup, I was curious what it looked like in the day. After hearing from
everybody, I will use vi the first time and if I have to do it a bunch
of ties, will test a sed script... or just wait 'til Warner's done and
just check it from source code :).
Will
--
GPG Fingerprint: 68F4 B3BD 1730 555A 4462 7D45 3EAA 5B6D A982 BAAF
I'm just getting comfortable using sed,
thanks for the
simplification.
The issue with sed is that different shar programs used different
cut/here markers. Using vi was safer and as I said, since you did not
do it everyday, it was no big deal.
Maybe since you have a number of these you want to apply, check with a
grep for the expected marker before you try to apply the sed script
and only do it in a more automated matter catching the vast majority
of the markers.]
You asked how we did it. As was said, there were a number of
unshar(1) commands out there, that did the sed for you. Go looking in
the archives, but generally speaking using vi was 'good enough' for
day to day work because you did not run it that often / every day /
were >>usually<< not apply a lot of them in a row.
Clem 
25 Jul
25 Jul
2 a.m.
On Fri, 24 Jul 2020, Random832 wrote:
For whatever it's worth, you can do the exact
same thing as vi with sed
in this case: 1,/====/d
It's been a while since I had to use it, but didn't "unshar" do this
sort
of thing, and in a safe manner too? You'd be out of your mind to blindly
run the shell on some anonymous shar file...
-- Dave
20 Jul
20 Jul
5:55 p.m.
On Mon, Jul 20, 2020 at 11:25 AM Will Senn <will.senn(a)gmail.com> wrote:
As y'all know, I'm a relative latecomer to
the world of Unix, but I do try
to figure out how y'all did it back when. So, sometimes, as in this case, I
can figure out how to do something, but I'm curious how it was done back in
the day, moreso than how I can get it done today. I'm looking at the
patching of my shiny new 2.11 BSD pl 431 system running on my speedy little
virtual PDP-11/70, so I grabbed patch 432 (here's a portion of the patch):
...
To install the update cut where indicated below and save to a file
(/tmp/432) and then:
cd /tmp
sh 432
./432.sh
./432.rm
sh 432.shar
patch -p0 < 432.patch
Watch carefully for any rejected parts of the patch. Failure of a
patch typically means the system was not current on all preceeding
updates _or_ that local modifications have been made.
...
====== cut here
#! /bin/sh
# This is a shell archive, meaning:
# 1. Remove everything above the #! /bin/sh line.
# 2. Save the resulting text in a file.
# 3. Execute the file with /bin/sh (not csh) to create:
# 432.rm
# 432.sh
# 432.shar
# 432.patch
...
# End of shell archive
This seems straightforward. Just follow the directions et voila magic
happens.
My questions for y'all are how would you go about doing this? Use vi to
delete everything through the ==== cut here line? Use some combination of
standard unix tools to chop it up? What was your workflow for patching up
the system using these files?
sed -e '1,/---cut here---/d' < $patch | sh -x
is what I use, but there's a wide variety of 'cut here' lines in the
2.11BSD patches, so I have had to taylor to each patch.
In my world, if I screw something up, it's 15
seconds to run a restore
script in my simh directory and I can try again, so my level of concern for
a mistake is pretty low. If I was doing this in 1980, on real hardware, I
would have had many concerns, as I'm sure some of y'all can remember, how
did you prepare and protect yourselves so a patch was successful.
Yea, it was always a crap-shoot back in the day on slow hardware. Backups
on tape were your best bet :(.
BTW, I thought .shar was an archive format, so when I
saw the patch was a
shar file, I was worried it would be in some binary form, lo and behold, it
looks like text to me... not even b64. So much to learn, so little time.
It is and it isn't. Mostly isn't for these patches. libarchive supports it,
but there's no standard and what libarchive supports is quite limited.
Warner
Thanks,
Will
--
GPG Fingerprint: 68F4 B3BD 1730 555A 4462 7D45 3EAA 5B6D A982 BAAF
6:11 p.m.
On 7/20/20 12:55 PM, Warner Losh wrote:
Will said:
My questions for y'all are how would you go about doing this? Use
vi to delete everything through the ==== cut here line? Use some
combination of standard unix tools to chop it up? What was your
workflow for patching up the system using these files?
sed -e '1,/---cut here---/d' < $patch | sh -x
is what I use, but there's a wide variety of 'cut here' lines in the
2.11BSD patches, so I have had to taylor to each patch.
Thanks, this is nice and
simple. I'll be on the lookout for the differences.
Will
6:41 p.m.
On Mon, Jul 20, 2020 at 2:12 PM Will Senn <will.senn(a)gmail.com> wrote:
sed -e '1,/---cut here---/d' < $patch |
sh -x
is what I use, but there's a wide variety of 'cut here' lines in the
2.11BSD patches, so I have had to taylor to each patch.
Yep and you usually only running one or two at a time. Your work As
Wernflow of
a applying a lot of patches is a tad different. As Warner
said, there were plenty of different shar/unshars back in the day.
Just remember to thank Mary Ann for uuencode/decode which started it all ;-)
Clem
1369
days inactive
1374
days old
13 comments
6 participants
participants (6)
-
Clem Cole -
Dave Horsfall -
Random832 -
Richard Salz -
Warner Losh -
Will Senn