Kill the SIGPIPE signal handler which is installed around write opertations.
Instead just SIG_IGN SIGPIPE in main.c for all of acme-client.
More work to be done here but at least this distraction is gone.
OK florian@ deraadt@ op@

Check that the challenge token which is turned into a filename is
base64url encoded.
We have only the challenge directory unveil(2)'ed so funny business
like ../ will not work, but we shouldn't generate garbage filenames
that someone else might trip over either.
Pointed out and diff by Ali Farzanrad (ali_farzanrad AT riseup.net)
OK beck

Allow to have multiple domain ... {} sextions with the same domain
name, by adding a new (optional) config option "domain name".
This can be used to create a rsa and an ecdsa key for the same domain
name.
The old domain name in the 'title' line continues to be used as domain
name in the abscence of the domain name argument, i.e. the change is
backward compatible with current config files.
tested by sthen@
ok florian@ sthen@

Do not define variables in extern.h since this will lead to duplicate
definitions in every source file that includes extern.h.
From Michael Forney (mforney AT mforney DOT org), thanks!
OK jca

Implement elliptic curve account keys.
OK benno
Input & OK tb

Trade unveil(2) for chroot(2).
This uses less code and unveil(2) seems to be the better tool here.
The directory one chroots into needs to be carefully setup (they are
not) and comon wisedom is that root can break out of chroots.
There is probably nothing wrong with the chroot code because of pledge
but it still makes me feel uneasy.
input & OK on previous version mestre
OK on previous version deraadt
bug found, input & OK benno

acctkey indirection is unnecessary, just pass authority->account

Track key type (RSA or ECDSA) in an enum and clean up a bit while here.
Originaly from Renaud Allard following input from benno, tweaked by me.
OK benno

use acme-client to sign certificated with ecdsa keys

diff from Renaud Allard <renaud@allard.it>, ok to get in from florian@

Remove A and D flag, they are superfluous.
One could always use them on the command line and acme-client would do
the right thing.

Implement RFC 8555 "Automatic Certificate Management Environment
(ACME)" to be able to talk to the v02 Let's Encrypt API.

With this acme-client(1) will no longer be able to talk to the v01
API. Users must change the api url in /etc/acme-client.conf to
https://acme-v02.api.letsencrypt.org/directory
Existing accounts (and certs of course) stay valid and after the url
change acme-client will be able to renew certs.

Tested by Renaud Allard and benno
Input & OK benno

remove free() before exit() from main(), in ongoing work to make
things more readable. otto notes that free() does some checks, but
in this case readability is better than complete cleanup.
ok florian@ deraadt@

use temporary strings for calls to basename(), because it may change
the contents of its argument (on other platforms). Also strdup() the result,
because basename returns a pointer to static memory.
ok florian@

fix dirname usage in acme-client: dirname() can modify its argument
(it does not on OpenBSD) so we need to us a copy of the string. In
addition, copy the result of dirname() as well, because it's static
storage and if we call dirname() again it will be overwritten.
Original problem noted and fix suggested by Wolf (wolf AT wolfsden DOT
cz)
ok florian@

add an extra check for existing "api ..." line in the config.
move the check of existing "account ..." line from main.c to the parser.
ok deraadt@

cleanup initialization of chngdir. inspired by Ross L Richardson.
ok tb@

replace exit() with return(), from Ross L Richardson.
ok tb@ (previous 3 commits to main.c as well)

replace warn() + exit() with err()
From Ross L Richardson.

line too long and whitespace. From Ross L Richardson.

reorder option parsing to be alphabetical/same order as usage and
manpage. From Ross L Richardson.

Deprecate agreement url config option and get the information from the
directory call. This way we don't need to update the acme-client.conf
file every time it changes. Still parse the option, ignore and warn about
it for a release. Sysmerge should be able to handle the removal.
"nice" deraadt@
OK benno

Use basename(3) instead of hand rolling it.
Gets rid of double slashes, pointed out by Raf Czlonka (rczlonka at
gmail), thanks!

OK benno

pull root check up, the parser will bomb out anyway
OK benno

Make returns consistent and remove parentheses per style(9).

Transformed with coccinelle.

Requested by and ok tb@

Complete jsing's coccinelle cleanup... by hand
ok jsing

errx() does not need extra newline

Replace comparisons between a constant or enum and an expression, with
a comparison between the expression and the constant or enum. This
significantly improves readability.

Transformed with coccinelle.

Requested by deraadt@

tweak previous;

move config data structures mostly to TAILQs, because that way we
preserve order.

ok florian@

it is ok to only specify a full chain certificate
OK benno

add option 'domain full chain certificate "path"',
revokation works, the fullchain file will be unlinked.

ok florian

Implement domain chain certificate.
OK benno

Split certificate file from config file into certdir and certfile.
This way we can still chroot to certdir but the the certificate file
is not fixed to "cert.pem".

Writing of chain.pem and fullchain.pem is currently broken with this.

OK benno

Remove backup option. This is not acme-client's business; also it gets
in the way.
OK benno

We are only dealing with one domain on the command line
OK benno

typo; ok benno

typo; ok benno

acme-client use configuration file [5 of 5]
implement new -n option to check and print configuration

ok florian

acme-client use configuration file [4 of 5]

fix getopt()

ok florian

acme-client use configuration file [3 of 5]

change command line options:

n -> A new Account key
N -> D new Domain key

With this acme-client has these main usage patterns:

* create new Account Key and Domain Key and get a certificate:

acme-client -A -D www.example.com

* renew certificate:

acme-client www.example.com

* revoke certificate:

acme-client -r www.example.com

ok florian

acme-client use configuration file [2 of 5]

- add challengedir option to config file
- remove -C option from command line

ok florian

acme-client use configuration file [1 of 5]

start using the configuration file and delete command line arguments:

-a agreement -> agreement url ...
-c certdir -> domain certificate "path"
-f accountkey -> account key "path"
-k domainkey -> domain key "path"
-s authority -> sign with "name"

new argument:
-f configfile

the changes needed to use the new configuration are local to main.c for now.
While the configuration could be passed directly to netproc(), keyproc() etc,
the diff is smaller this way.

This also removes the multidir (-m) mode for now - specify different paths in
each domain {} block instead.

ok florian

add a config file parser to acme-client (unused at the moment, so that
it can be worked on in the tree).
ok florian@ deraadt@

A small amount more KNF to make this easier on our eyes. Not going further
than this.
ok florian

hoist local variable initialization directly into the definitions,
rather than doing it right afterwards.
ok florian

Inconsistant use of case (value): idiom drives me nuts, I have no idea
what purpose it was intended to serve.

update usage(); pointed out by jmc@

Implement table driven selection to which ACME authorities we can
talk.
Suggest by and OK deraadt, OK benno.

(Later on deraadt and benno discussed if this should be handled with a
config file. This seems to be good enough for now. We can do a config
file later.)

we don't have config.h

line wrap usage

Collapse krazy abstractions for other sandbox models, using chroot and
pledge directly as needed.
ok florian

knf

whitespace

revoke -> revocate, to avoid alias
ok florian

store files in directories called acme

Import Kristaps' letskencrypt and call it acme-client in tree.
OK to get it in deraadt@ (and probably beck@)

At least deraadt@, beck@ and otto@ are fine with the name and the
disagreements stopped.

Check that the challenge token which is turned into a filename is
base64url encoded.
We have only the challenge directory unveil(2)'ed so funny business
like ../ will not work, but we shouldn't generate garbage filenames
that someone else might trip over either.
Pointed out and diff by Ali Farzanrad (ali_farzanrad AT riseup.net)
OK beck

Allow to have multiple domain ... {} sextions with the same domain
name, by adding a new (optional) config option "domain name".
This can be used to create a rsa and an ecdsa key for the same domain
name.
The old domain name in the 'title' line continues to be used as domain
name in the abscence of the domain name argument, i.e. the change is
backward compatible with current config files.
tested by sthen@
ok florian@ sthen@

Do not define variables in extern.h since this will lead to duplicate
definitions in every source file that includes extern.h.
From Michael Forney (mforney AT mforney DOT org), thanks!
OK jca

Implement elliptic curve account keys.
OK benno
Input & OK tb

Trade unveil(2) for chroot(2).
This uses less code and unveil(2) seems to be the better tool here.
The directory one chroots into needs to be carefully setup (they are
not) and comon wisedom is that root can break out of chroots.
There is probably nothing wrong with the chroot code because of pledge
but it still makes me feel uneasy.
input & OK on previous version mestre
OK on previous version deraadt
bug found, input & OK benno

acctkey indirection is unnecessary, just pass authority->account

Track key type (RSA or ECDSA) in an enum and clean up a bit while here.
Originaly from Renaud Allard following input from benno, tweaked by me.
OK benno

use acme-client to sign certificated with ecdsa keys

diff from Renaud Allard <renaud@allard.it>, ok to get in from florian@

Remove A and D flag, they are superfluous.
One could always use them on the command line and acme-client would do
the right thing.

Implement RFC 8555 "Automatic Certificate Management Environment
(ACME)" to be able to talk to the v02 Let's Encrypt API.

With this acme-client(1) will no longer be able to talk to the v01
API. Users must change the api url in /etc/acme-client.conf to
https://acme-v02.api.letsencrypt.org/directory
Existing accounts (and certs of course) stay valid and after the url
change acme-client will be able to renew certs.

Tested by Renaud Allard and benno
Input & OK benno

remove free() before exit() from main(), in ongoing work to make
things more readable. otto notes that free() does some checks, but
in this case readability is better than complete cleanup.
ok florian@ deraadt@

use temporary strings for calls to basename(), because it may change
the contents of its argument (on other platforms). Also strdup() the result,
because basename returns a pointer to static memory.
ok florian@

fix dirname usage in acme-client: dirname() can modify its argument
(it does not on OpenBSD) so we need to us a copy of the string. In
addition, copy the result of dirname() as well, because it's static
storage and if we call dirname() again it will be overwritten.
Original problem noted and fix suggested by Wolf (wolf AT wolfsden DOT
cz)
ok florian@

add an extra check for existing "api ..." line in the config.
move the check of existing "account ..." line from main.c to the parser.
ok deraadt@

cleanup initialization of chngdir. inspired by Ross L Richardson.
ok tb@

replace exit() with return(), from Ross L Richardson.
ok tb@ (previous 3 commits to main.c as well)

replace warn() + exit() with err()
From Ross L Richardson.

line too long and whitespace. From Ross L Richardson.

reorder option parsing to be alphabetical/same order as usage and
manpage. From Ross L Richardson.

Deprecate agreement url config option and get the information from the
directory call. This way we don't need to update the acme-client.conf
file every time it changes. Still parse the option, ignore and warn about
it for a release. Sysmerge should be able to handle the removal.
"nice" deraadt@
OK benno

Use basename(3) instead of hand rolling it.
Gets rid of double slashes, pointed out by Raf Czlonka (rczlonka at
gmail), thanks!

OK benno

pull root check up, the parser will bomb out anyway
OK benno

Make returns consistent and remove parentheses per style(9).

Transformed with coccinelle.

Requested by and ok tb@

Complete jsing's coccinelle cleanup... by hand
ok jsing

errx() does not need extra newline

Replace comparisons between a constant or enum and an expression, with
a comparison between the expression and the constant or enum. This
significantly improves readability.

Transformed with coccinelle.

Requested by deraadt@

tweak previous;

move config data structures mostly to TAILQs, because that way we
preserve order.

ok florian@

it is ok to only specify a full chain certificate
OK benno

add option 'domain full chain certificate "path"',
revokation works, the fullchain file will be unlinked.

ok florian

Implement domain chain certificate.
OK benno

Split certificate file from config file into certdir and certfile.
This way we can still chroot to certdir but the the certificate file
is not fixed to "cert.pem".

Writing of chain.pem and fullchain.pem is currently broken with this.

OK benno

Remove backup option. This is not acme-client's business; also it gets
in the way.
OK benno

We are only dealing with one domain on the command line
OK benno

typo; ok benno

typo; ok benno

acme-client use configuration file [5 of 5]
implement new -n option to check and print configuration

ok florian

acme-client use configuration file [4 of 5]

fix getopt()

ok florian

acme-client use configuration file [3 of 5]

change command line options:

n -> A new Account key
N -> D new Domain key

With this acme-client has these main usage patterns:

* create new Account Key and Domain Key and get a certificate:

acme-client -A -D www.example.com

* renew certificate:

acme-client www.example.com

* revoke certificate:

acme-client -r www.example.com

ok florian

acme-client use configuration file [2 of 5]

- add challengedir option to config file
- remove -C option from command line

ok florian

acme-client use configuration file [1 of 5]

start using the configuration file and delete command line arguments:

-a agreement -> agreement url ...
-c certdir -> domain certificate "path"
-f accountkey -> account key "path"
-k domainkey -> domain key "path"
-s authority -> sign with "name"

new argument:
-f configfile

the changes needed to use the new configuration are local to main.c for now.
While the configuration could be passed directly to netproc(), keyproc() etc,
the diff is smaller this way.

This also removes the multidir (-m) mode for now - specify different paths in
each domain {} block instead.

ok florian

add a config file parser to acme-client (unused at the moment, so that
it can be worked on in the tree).
ok florian@ deraadt@

A small amount more KNF to make this easier on our eyes. Not going further
than this.
ok florian

hoist local variable initialization directly into the definitions,
rather than doing it right afterwards.
ok florian

Inconsistant use of case (value): idiom drives me nuts, I have no idea
what purpose it was intended to serve.

update usage(); pointed out by jmc@

Implement table driven selection to which ACME authorities we can
talk.
Suggest by and OK deraadt, OK benno.

(Later on deraadt and benno discussed if this should be handled with a
config file. This seems to be good enough for now. We can do a config
file later.)

we don't have config.h

line wrap usage

Collapse krazy abstractions for other sandbox models, using chroot and
pledge directly as needed.
ok florian

knf

whitespace

revoke -> revocate, to avoid alias
ok florian

store files in directories called acme

Import Kristaps' letskencrypt and call it acme-client in tree.
OK to get it in deraadt@ (and probably beck@)

At least deraadt@, beck@ and otto@ are fine with the name and the
disagreements stopped.

Allow to have multiple domain ... {} sextions with the same domain
name, by adding a new (optional) config option "domain name".
This can be used to create a rsa and an ecdsa key for the same domain
name.
The old domain name in the 'title' line continues to be used as domain
name in the abscence of the domain name argument, i.e. the change is
backward compatible with current config files.
tested by sthen@
ok florian@ sthen@

Do not define variables in extern.h since this will lead to duplicate
definitions in every source file that includes extern.h.
From Michael Forney (mforney AT mforney DOT org), thanks!
OK jca

Implement elliptic curve account keys.
OK benno
Input & OK tb

Trade unveil(2) for chroot(2).
This uses less code and unveil(2) seems to be the better tool here.
The directory one chroots into needs to be carefully setup (they are
not) and comon wisedom is that root can break out of chroots.
There is probably nothing wrong with the chroot code because of pledge
but it still makes me feel uneasy.
input & OK on previous version mestre
OK on previous version deraadt
bug found, input & OK benno

acctkey indirection is unnecessary, just pass authority->account

Track key type (RSA or ECDSA) in an enum and clean up a bit while here.
Originaly from Renaud Allard following input from benno, tweaked by me.
OK benno

use acme-client to sign certificated with ecdsa keys

diff from Renaud Allard <renaud@allard.it>, ok to get in from florian@

Remove A and D flag, they are superfluous.
One could always use them on the command line and acme-client would do
the right thing.

Implement RFC 8555 "Automatic Certificate Management Environment
(ACME)" to be able to talk to the v02 Let's Encrypt API.

With this acme-client(1) will no longer be able to talk to the v01
API. Users must change the api url in /etc/acme-client.conf to
https://acme-v02.api.letsencrypt.org/directory
Existing accounts (and certs of course) stay valid and after the url
change acme-client will be able to renew certs.

Tested by Renaud Allard and benno
Input & OK benno

remove free() before exit() from main(), in ongoing work to make
things more readable. otto notes that free() does some checks, but
in this case readability is better than complete cleanup.
ok florian@ deraadt@

use temporary strings for calls to basename(), because it may change
the contents of its argument (on other platforms). Also strdup() the result,
because basename returns a pointer to static memory.
ok florian@

fix dirname usage in acme-client: dirname() can modify its argument
(it does not on OpenBSD) so we need to us a copy of the string. In
addition, copy the result of dirname() as well, because it's static
storage and if we call dirname() again it will be overwritten.
Original problem noted and fix suggested by Wolf (wolf AT wolfsden DOT
cz)
ok florian@

add an extra check for existing "api ..." line in the config.
move the check of existing "account ..." line from main.c to the parser.
ok deraadt@

cleanup initialization of chngdir. inspired by Ross L Richardson.
ok tb@

replace exit() with return(), from Ross L Richardson.
ok tb@ (previous 3 commits to main.c as well)

replace warn() + exit() with err()
From Ross L Richardson.

line too long and whitespace. From Ross L Richardson.

reorder option parsing to be alphabetical/same order as usage and
manpage. From Ross L Richardson.

Deprecate agreement url config option and get the information from the
directory call. This way we don't need to update the acme-client.conf
file every time it changes. Still parse the option, ignore and warn about
it for a release. Sysmerge should be able to handle the removal.
"nice" deraadt@
OK benno

Use basename(3) instead of hand rolling it.
Gets rid of double slashes, pointed out by Raf Czlonka (rczlonka at
gmail), thanks!

OK benno

pull root check up, the parser will bomb out anyway
OK benno

Make returns consistent and remove parentheses per style(9).

Transformed with coccinelle.

Requested by and ok tb@

Complete jsing's coccinelle cleanup... by hand
ok jsing

errx() does not need extra newline

Replace comparisons between a constant or enum and an expression, with
a comparison between the expression and the constant or enum. This
significantly improves readability.

Transformed with coccinelle.

Requested by deraadt@

tweak previous;

move config data structures mostly to TAILQs, because that way we
preserve order.

ok florian@

it is ok to only specify a full chain certificate
OK benno

add option 'domain full chain certificate "path"',
revokation works, the fullchain file will be unlinked.

ok florian

Implement domain chain certificate.
OK benno

Split certificate file from config file into certdir and certfile.
This way we can still chroot to certdir but the the certificate file
is not fixed to "cert.pem".

Writing of chain.pem and fullchain.pem is currently broken with this.

OK benno

Remove backup option. This is not acme-client's business; also it gets
in the way.
OK benno

We are only dealing with one domain on the command line
OK benno

typo; ok benno

typo; ok benno

acme-client use configuration file [5 of 5]
implement new -n option to check and print configuration

ok florian

acme-client use configuration file [4 of 5]

fix getopt()

ok florian

acme-client use configuration file [3 of 5]

change command line options:

n -> A new Account key
N -> D new Domain key

With this acme-client has these main usage patterns:

* create new Account Key and Domain Key and get a certificate:

acme-client -A -D www.example.com

* renew certificate:

acme-client www.example.com

* revoke certificate:

acme-client -r www.example.com

ok florian

acme-client use configuration file [2 of 5]

- add challengedir option to config file
- remove -C option from command line

ok florian

acme-client use configuration file [1 of 5]

start using the configuration file and delete command line arguments:

-a agreement -> agreement url ...
-c certdir -> domain certificate "path"
-f accountkey -> account key "path"
-k domainkey -> domain key "path"
-s authority -> sign with "name"

new argument:
-f configfile

the changes needed to use the new configuration are local to main.c for now.
While the configuration could be passed directly to netproc(), keyproc() etc,
the diff is smaller this way.

This also removes the multidir (-m) mode for now - specify different paths in
each domain {} block instead.

ok florian

add a config file parser to acme-client (unused at the moment, so that
it can be worked on in the tree).
ok florian@ deraadt@

A small amount more KNF to make this easier on our eyes. Not going further
than this.
ok florian

hoist local variable initialization directly into the definitions,
rather than doing it right afterwards.
ok florian

Inconsistant use of case (value): idiom drives me nuts, I have no idea
what purpose it was intended to serve.

update usage(); pointed out by jmc@

Implement table driven selection to which ACME authorities we can
talk.
Suggest by and OK deraadt, OK benno.

(Later on deraadt and benno discussed if this should be handled with a
config file. This seems to be good enough for now. We can do a config
file later.)

we don't have config.h

line wrap usage

Collapse krazy abstractions for other sandbox models, using chroot and
pledge directly as needed.
ok florian

knf

whitespace

revoke -> revocate, to avoid alias
ok florian

store files in directories called acme

Import Kristaps' letskencrypt and call it acme-client in tree.
OK to get it in deraadt@ (and probably beck@)

At least deraadt@, beck@ and otto@ are fine with the name and the
disagreements stopped.

Do not define variables in extern.h since this will lead to duplicate
definitions in every source file that includes extern.h.
From Michael Forney (mforney AT mforney DOT org), thanks!
OK jca

Implement elliptic curve account keys.
OK benno
Input & OK tb

Trade unveil(2) for chroot(2).
This uses less code and unveil(2) seems to be the better tool here.
The directory one chroots into needs to be carefully setup (they are
not) and comon wisedom is that root can break out of chroots.
There is probably nothing wrong with the chroot code because of pledge
but it still makes me feel uneasy.
input & OK on previous version mestre
OK on previous version deraadt
bug found, input & OK benno

acctkey indirection is unnecessary, just pass authority->account

Track key type (RSA or ECDSA) in an enum and clean up a bit while here.
Originaly from Renaud Allard following input from benno, tweaked by me.
OK benno

use acme-client to sign certificated with ecdsa keys

diff from Renaud Allard <renaud@allard.it>, ok to get in from florian@

Remove A and D flag, they are superfluous.
One could always use them on the command line and acme-client would do
the right thing.

Implement RFC 8555 "Automatic Certificate Management Environment
(ACME)" to be able to talk to the v02 Let's Encrypt API.

With this acme-client(1) will no longer be able to talk to the v01
API. Users must change the api url in /etc/acme-client.conf to
https://acme-v02.api.letsencrypt.org/directory
Existing accounts (and certs of course) stay valid and after the url
change acme-client will be able to renew certs.

Tested by Renaud Allard and benno
Input & OK benno

remove free() before exit() from main(), in ongoing work to make
things more readable. otto notes that free() does some checks, but
in this case readability is better than complete cleanup.
ok florian@ deraadt@

use temporary strings for calls to basename(), because it may change
the contents of its argument (on other platforms). Also strdup() the result,
because basename returns a pointer to static memory.
ok florian@

fix dirname usage in acme-client: dirname() can modify its argument
(it does not on OpenBSD) so we need to us a copy of the string. In
addition, copy the result of dirname() as well, because it's static
storage and if we call dirname() again it will be overwritten.
Original problem noted and fix suggested by Wolf (wolf AT wolfsden DOT
cz)
ok florian@

add an extra check for existing "api ..." line in the config.
move the check of existing "account ..." line from main.c to the parser.
ok deraadt@

cleanup initialization of chngdir. inspired by Ross L Richardson.
ok tb@

replace exit() with return(), from Ross L Richardson.
ok tb@ (previous 3 commits to main.c as well)

replace warn() + exit() with err()
From Ross L Richardson.

line too long and whitespace. From Ross L Richardson.

reorder option parsing to be alphabetical/same order as usage and
manpage. From Ross L Richardson.

Deprecate agreement url config option and get the information from the
directory call. This way we don't need to update the acme-client.conf
file every time it changes. Still parse the option, ignore and warn about
it for a release. Sysmerge should be able to handle the removal.
"nice" deraadt@
OK benno

Use basename(3) instead of hand rolling it.
Gets rid of double slashes, pointed out by Raf Czlonka (rczlonka at
gmail), thanks!

OK benno

pull root check up, the parser will bomb out anyway
OK benno

Make returns consistent and remove parentheses per style(9).

Transformed with coccinelle.

Requested by and ok tb@

Complete jsing's coccinelle cleanup... by hand
ok jsing

errx() does not need extra newline

Replace comparisons between a constant or enum and an expression, with
a comparison between the expression and the constant or enum. This
significantly improves readability.

Transformed with coccinelle.

Requested by deraadt@

tweak previous;

move config data structures mostly to TAILQs, because that way we
preserve order.

ok florian@

it is ok to only specify a full chain certificate
OK benno

add option 'domain full chain certificate "path"',
revokation works, the fullchain file will be unlinked.

ok florian

Implement domain chain certificate.
OK benno

Split certificate file from config file into certdir and certfile.
This way we can still chroot to certdir but the the certificate file
is not fixed to "cert.pem".

Writing of chain.pem and fullchain.pem is currently broken with this.

OK benno

Remove backup option. This is not acme-client's business; also it gets
in the way.
OK benno

We are only dealing with one domain on the command line
OK benno

typo; ok benno

typo; ok benno

acme-client use configuration file [5 of 5]
implement new -n option to check and print configuration

ok florian

acme-client use configuration file [4 of 5]

fix getopt()

ok florian

acme-client use configuration file [3 of 5]

change command line options:

n -> A new Account key
N -> D new Domain key

With this acme-client has these main usage patterns:

* create new Account Key and Domain Key and get a certificate:

acme-client -A -D www.example.com

* renew certificate:

acme-client www.example.com

* revoke certificate:

acme-client -r www.example.com

ok florian

acme-client use configuration file [2 of 5]

- add challengedir option to config file
- remove -C option from command line

ok florian

acme-client use configuration file [1 of 5]

start using the configuration file and delete command line arguments:

-a agreement -> agreement url ...
-c certdir -> domain certificate "path"
-f accountkey -> account key "path"
-k domainkey -> domain key "path"
-s authority -> sign with "name"

new argument:
-f configfile

the changes needed to use the new configuration are local to main.c for now.
While the configuration could be passed directly to netproc(), keyproc() etc,
the diff is smaller this way.

This also removes the multidir (-m) mode for now - specify different paths in
each domain {} block instead.

ok florian

add a config file parser to acme-client (unused at the moment, so that
it can be worked on in the tree).
ok florian@ deraadt@

A small amount more KNF to make this easier on our eyes. Not going further
than this.
ok florian

hoist local variable initialization directly into the definitions,
rather than doing it right afterwards.
ok florian

Inconsistant use of case (value): idiom drives me nuts, I have no idea
what purpose it was intended to serve.

update usage(); pointed out by jmc@

Implement table driven selection to which ACME authorities we can
talk.
Suggest by and OK deraadt, OK benno.

(Later on deraadt and benno discussed if this should be handled with a
config file. This seems to be good enough for now. We can do a config
file later.)

we don't have config.h

line wrap usage

Collapse krazy abstractions for other sandbox models, using chroot and
pledge directly as needed.
ok florian

knf

whitespace

revoke -> revocate, to avoid alias
ok florian

store files in directories called acme

Import Kristaps' letskencrypt and call it acme-client in tree.
OK to get it in deraadt@ (and probably beck@)

At least deraadt@, beck@ and otto@ are fine with the name and the
disagreements stopped.

Implement elliptic curve account keys.
OK benno
Input & OK tb

Trade unveil(2) for chroot(2).
This uses less code and unveil(2) seems to be the better tool here.
The directory one chroots into needs to be carefully setup (they are
not) and comon wisedom is that root can break out of chroots.
There is probably nothing wrong with the chroot code because of pledge
but it still makes me feel uneasy.
input & OK on previous version mestre
OK on previous version deraadt
bug found, input & OK benno

acctkey indirection is unnecessary, just pass authority->account

Track key type (RSA or ECDSA) in an enum and clean up a bit while here.
Originaly from Renaud Allard following input from benno, tweaked by me.
OK benno

use acme-client to sign certificated with ecdsa keys

diff from Renaud Allard <renaud@allard.it>, ok to get in from florian@

Remove A and D flag, they are superfluous.
One could always use them on the command line and acme-client would do
the right thing.

Implement RFC 8555 "Automatic Certificate Management Environment
(ACME)" to be able to talk to the v02 Let's Encrypt API.

With this acme-client(1) will no longer be able to talk to the v01
API. Users must change the api url in /etc/acme-client.conf to
https://acme-v02.api.letsencrypt.org/directory
Existing accounts (and certs of course) stay valid and after the url
change acme-client will be able to renew certs.

Tested by Renaud Allard and benno
Input & OK benno

remove free() before exit() from main(), in ongoing work to make
things more readable. otto notes that free() does some checks, but
in this case readability is better than complete cleanup.
ok florian@ deraadt@

use temporary strings for calls to basename(), because it may change
the contents of its argument (on other platforms). Also strdup() the result,
because basename returns a pointer to static memory.
ok florian@

fix dirname usage in acme-client: dirname() can modify its argument
(it does not on OpenBSD) so we need to us a copy of the string. In
addition, copy the result of dirname() as well, because it's static
storage and if we call dirname() again it will be overwritten.
Original problem noted and fix suggested by Wolf (wolf AT wolfsden DOT
cz)
ok florian@

add an extra check for existing "api ..." line in the config.
move the check of existing "account ..." line from main.c to the parser.
ok deraadt@

cleanup initialization of chngdir. inspired by Ross L Richardson.
ok tb@

replace exit() with return(), from Ross L Richardson.
ok tb@ (previous 3 commits to main.c as well)

replace warn() + exit() with err()
From Ross L Richardson.

line too long and whitespace. From Ross L Richardson.

reorder option parsing to be alphabetical/same order as usage and
manpage. From Ross L Richardson.

Deprecate agreement url config option and get the information from the
directory call. This way we don't need to update the acme-client.conf
file every time it changes. Still parse the option, ignore and warn about
it for a release. Sysmerge should be able to handle the removal.
"nice" deraadt@
OK benno

Use basename(3) instead of hand rolling it.
Gets rid of double slashes, pointed out by Raf Czlonka (rczlonka at
gmail), thanks!

OK benno

pull root check up, the parser will bomb out anyway
OK benno

Make returns consistent and remove parentheses per style(9).

Transformed with coccinelle.

Requested by and ok tb@

Complete jsing's coccinelle cleanup... by hand
ok jsing

errx() does not need extra newline

Replace comparisons between a constant or enum and an expression, with
a comparison between the expression and the constant or enum. This
significantly improves readability.

Transformed with coccinelle.

Requested by deraadt@

tweak previous;

move config data structures mostly to TAILQs, because that way we
preserve order.

ok florian@

it is ok to only specify a full chain certificate
OK benno

add option 'domain full chain certificate "path"',
revokation works, the fullchain file will be unlinked.

ok florian

Implement domain chain certificate.
OK benno

Split certificate file from config file into certdir and certfile.
This way we can still chroot to certdir but the the certificate file
is not fixed to "cert.pem".

Writing of chain.pem and fullchain.pem is currently broken with this.

OK benno

Remove backup option. This is not acme-client's business; also it gets
in the way.
OK benno

We are only dealing with one domain on the command line
OK benno

typo; ok benno

typo; ok benno

acme-client use configuration file [5 of 5]
implement new -n option to check and print configuration

ok florian

acme-client use configuration file [4 of 5]

fix getopt()

ok florian

acme-client use configuration file [3 of 5]

change command line options:

n -> A new Account key
N -> D new Domain key

With this acme-client has these main usage patterns:

* create new Account Key and Domain Key and get a certificate:

acme-client -A -D www.example.com

* renew certificate:

acme-client www.example.com

* revoke certificate:

acme-client -r www.example.com

ok florian

acme-client use configuration file [2 of 5]

- add challengedir option to config file
- remove -C option from command line

ok florian

acme-client use configuration file [1 of 5]

start using the configuration file and delete command line arguments:

-a agreement -> agreement url ...
-c certdir -> domain certificate "path"
-f accountkey -> account key "path"
-k domainkey -> domain key "path"
-s authority -> sign with "name"

new argument:
-f configfile

the changes needed to use the new configuration are local to main.c for now.
While the configuration could be passed directly to netproc(), keyproc() etc,
the diff is smaller this way.

This also removes the multidir (-m) mode for now - specify different paths in
each domain {} block instead.

ok florian

add a config file parser to acme-client (unused at the moment, so that
it can be worked on in the tree).
ok florian@ deraadt@

A small amount more KNF to make this easier on our eyes. Not going further
than this.
ok florian

hoist local variable initialization directly into the definitions,
rather than doing it right afterwards.
ok florian

Inconsistant use of case (value): idiom drives me nuts, I have no idea
what purpose it was intended to serve.

update usage(); pointed out by jmc@

Implement table driven selection to which ACME authorities we can
talk.
Suggest by and OK deraadt, OK benno.

(Later on deraadt and benno discussed if this should be handled with a
config file. This seems to be good enough for now. We can do a config
file later.)

we don't have config.h

line wrap usage

Collapse krazy abstractions for other sandbox models, using chroot and
pledge directly as needed.
ok florian

knf

whitespace

revoke -> revocate, to avoid alias
ok florian

store files in directories called acme

Import Kristaps' letskencrypt and call it acme-client in tree.
OK to get it in deraadt@ (and probably beck@)

At least deraadt@, beck@ and otto@ are fine with the name and the
disagreements stopped.

Trade unveil(2) for chroot(2).
This uses less code and unveil(2) seems to be the better tool here.
The directory one chroots into needs to be carefully setup (they are
not) and comon wisedom is that root can break out of chroots.
There is probably nothing wrong with the chroot code because of pledge
but it still makes me feel uneasy.
input & OK on previous version mestre
OK on previous version deraadt
bug found, input & OK benno

acctkey indirection is unnecessary, just pass authority->account

Track key type (RSA or ECDSA) in an enum and clean up a bit while here.
Originaly from Renaud Allard following input from benno, tweaked by me.
OK benno

use acme-client to sign certificated with ecdsa keys

diff from Renaud Allard <renaud@allard.it>, ok to get in from florian@

Remove A and D flag, they are superfluous.
One could always use them on the command line and acme-client would do
the right thing.

Implement RFC 8555 "Automatic Certificate Management Environment
(ACME)" to be able to talk to the v02 Let's Encrypt API.

With this acme-client(1) will no longer be able to talk to the v01
API. Users must change the api url in /etc/acme-client.conf to
https://acme-v02.api.letsencrypt.org/directory
Existing accounts (and certs of course) stay valid and after the url
change acme-client will be able to renew certs.

Tested by Renaud Allard and benno
Input & OK benno

remove free() before exit() from main(), in ongoing work to make
things more readable. otto notes that free() does some checks, but
in this case readability is better than complete cleanup.
ok florian@ deraadt@

use temporary strings for calls to basename(), because it may change
the contents of its argument (on other platforms). Also strdup() the result,
because basename returns a pointer to static memory.
ok florian@

fix dirname usage in acme-client: dirname() can modify its argument
(it does not on OpenBSD) so we need to us a copy of the string. In
addition, copy the result of dirname() as well, because it's static
storage and if we call dirname() again it will be overwritten.
Original problem noted and fix suggested by Wolf (wolf AT wolfsden DOT
cz)
ok florian@

add an extra check for existing "api ..." line in the config.
move the check of existing "account ..." line from main.c to the parser.
ok deraadt@

cleanup initialization of chngdir. inspired by Ross L Richardson.
ok tb@

replace exit() with return(), from Ross L Richardson.
ok tb@ (previous 3 commits to main.c as well)

replace warn() + exit() with err()
From Ross L Richardson.

line too long and whitespace. From Ross L Richardson.

reorder option parsing to be alphabetical/same order as usage and
manpage. From Ross L Richardson.

Deprecate agreement url config option and get the information from the
directory call. This way we don't need to update the acme-client.conf
file every time it changes. Still parse the option, ignore and warn about
it for a release. Sysmerge should be able to handle the removal.
"nice" deraadt@
OK benno

Use basename(3) instead of hand rolling it.
Gets rid of double slashes, pointed out by Raf Czlonka (rczlonka at
gmail), thanks!

OK benno

pull root check up, the parser will bomb out anyway
OK benno

Make returns consistent and remove parentheses per style(9).

Transformed with coccinelle.

Requested by and ok tb@

Complete jsing's coccinelle cleanup... by hand
ok jsing

errx() does not need extra newline

Replace comparisons between a constant or enum and an expression, with
a comparison between the expression and the constant or enum. This
significantly improves readability.

Transformed with coccinelle.

Requested by deraadt@

tweak previous;

move config data structures mostly to TAILQs, because that way we
preserve order.

ok florian@

it is ok to only specify a full chain certificate
OK benno

add option 'domain full chain certificate "path"',
revokation works, the fullchain file will be unlinked.

ok florian

Implement domain chain certificate.
OK benno

Split certificate file from config file into certdir and certfile.
This way we can still chroot to certdir but the the certificate file
is not fixed to "cert.pem".

Writing of chain.pem and fullchain.pem is currently broken with this.

OK benno

Remove backup option. This is not acme-client's business; also it gets
in the way.
OK benno

We are only dealing with one domain on the command line
OK benno

typo; ok benno

typo; ok benno

acme-client use configuration file [5 of 5]
implement new -n option to check and print configuration

ok florian

acme-client use configuration file [4 of 5]

fix getopt()

ok florian

acme-client use configuration file [3 of 5]

change command line options:

n -> A new Account key
N -> D new Domain key

With this acme-client has these main usage patterns:

* create new Account Key and Domain Key and get a certificate:

acme-client -A -D www.example.com

* renew certificate:

acme-client www.example.com

* revoke certificate:

acme-client -r www.example.com

ok florian

acme-client use configuration file [2 of 5]

- add challengedir option to config file
- remove -C option from command line

ok florian

acme-client use configuration file [1 of 5]

start using the configuration file and delete command line arguments:

-a agreement -> agreement url ...
-c certdir -> domain certificate "path"
-f accountkey -> account key "path"
-k domainkey -> domain key "path"
-s authority -> sign with "name"

new argument:
-f configfile

the changes needed to use the new configuration are local to main.c for now.
While the configuration could be passed directly to netproc(), keyproc() etc,
the diff is smaller this way.

This also removes the multidir (-m) mode for now - specify different paths in
each domain {} block instead.

ok florian

add a config file parser to acme-client (unused at the moment, so that
it can be worked on in the tree).
ok florian@ deraadt@

A small amount more KNF to make this easier on our eyes. Not going further
than this.
ok florian

hoist local variable initialization directly into the definitions,
rather than doing it right afterwards.
ok florian

Inconsistant use of case (value): idiom drives me nuts, I have no idea
what purpose it was intended to serve.

update usage(); pointed out by jmc@

Implement table driven selection to which ACME authorities we can
talk.
Suggest by and OK deraadt, OK benno.

(Later on deraadt and benno discussed if this should be handled with a
config file. This seems to be good enough for now. We can do a config
file later.)

we don't have config.h

line wrap usage

Collapse krazy abstractions for other sandbox models, using chroot and
pledge directly as needed.
ok florian

knf

whitespace

revoke -> revocate, to avoid alias
ok florian

store files in directories called acme

Import Kristaps' letskencrypt and call it acme-client in tree.
OK to get it in deraadt@ (and probably beck@)

At least deraadt@, beck@ and otto@ are fine with the name and the
disagreements stopped.

acctkey indirection is unnecessary, just pass authority->account

Track key type (RSA or ECDSA) in an enum and clean up a bit while here.
Originaly from Renaud Allard following input from benno, tweaked by me.
OK benno

use acme-client to sign certificated with ecdsa keys

diff from Renaud Allard <renaud@allard.it>, ok to get in from florian@

Remove A and D flag, they are superfluous.
One could always use them on the command line and acme-client would do
the right thing.

Implement RFC 8555 "Automatic Certificate Management Environment
(ACME)" to be able to talk to the v02 Let's Encrypt API.

With this acme-client(1) will no longer be able to talk to the v01
API. Users must change the api url in /etc/acme-client.conf to
https://acme-v02.api.letsencrypt.org/directory
Existing accounts (and certs of course) stay valid and after the url
change acme-client will be able to renew certs.

Tested by Renaud Allard and benno
Input & OK benno

remove free() before exit() from main(), in ongoing work to make
things more readable. otto notes that free() does some checks, but
in this case readability is better than complete cleanup.
ok florian@ deraadt@

use temporary strings for calls to basename(), because it may change
the contents of its argument (on other platforms). Also strdup() the result,
because basename returns a pointer to static memory.
ok florian@

fix dirname usage in acme-client: dirname() can modify its argument
(it does not on OpenBSD) so we need to us a copy of the string. In
addition, copy the result of dirname() as well, because it's static
storage and if we call dirname() again it will be overwritten.
Original problem noted and fix suggested by Wolf (wolf AT wolfsden DOT
cz)
ok florian@

add an extra check for existing "api ..." line in the config.
move the check of existing "account ..." line from main.c to the parser.
ok deraadt@

cleanup initialization of chngdir. inspired by Ross L Richardson.
ok tb@

replace exit() with return(), from Ross L Richardson.
ok tb@ (previous 3 commits to main.c as well)

replace warn() + exit() with err()
From Ross L Richardson.

line too long and whitespace. From Ross L Richardson.

reorder option parsing to be alphabetical/same order as usage and
manpage. From Ross L Richardson.

Deprecate agreement url config option and get the information from the
directory call. This way we don't need to update the acme-client.conf
file every time it changes. Still parse the option, ignore and warn about
it for a release. Sysmerge should be able to handle the removal.
"nice" deraadt@
OK benno

Use basename(3) instead of hand rolling it.
Gets rid of double slashes, pointed out by Raf Czlonka (rczlonka at
gmail), thanks!

OK benno

pull root check up, the parser will bomb out anyway
OK benno

Make returns consistent and remove parentheses per style(9).

Transformed with coccinelle.

Requested by and ok tb@

Complete jsing's coccinelle cleanup... by hand
ok jsing

errx() does not need extra newline

Replace comparisons between a constant or enum and an expression, with
a comparison between the expression and the constant or enum. This
significantly improves readability.

Transformed with coccinelle.

Requested by deraadt@

tweak previous;

move config data structures mostly to TAILQs, because that way we
preserve order.

ok florian@

it is ok to only specify a full chain certificate
OK benno

add option 'domain full chain certificate "path"',
revokation works, the fullchain file will be unlinked.

ok florian

Implement domain chain certificate.
OK benno

Split certificate file from config file into certdir and certfile.
This way we can still chroot to certdir but the the certificate file
is not fixed to "cert.pem".

Writing of chain.pem and fullchain.pem is currently broken with this.

OK benno

Remove backup option. This is not acme-client's business; also it gets
in the way.
OK benno

We are only dealing with one domain on the command line
OK benno

typo; ok benno

typo; ok benno

acme-client use configuration file [5 of 5]
implement new -n option to check and print configuration

ok florian

acme-client use configuration file [4 of 5]

fix getopt()

ok florian

acme-client use configuration file [3 of 5]

change command line options:

n -> A new Account key
N -> D new Domain key

With this acme-client has these main usage patterns:

* create new Account Key and Domain Key and get a certificate:

acme-client -A -D www.example.com

* renew certificate:

acme-client www.example.com

* revoke certificate:

acme-client -r www.example.com

ok florian

acme-client use configuration file [2 of 5]

- add challengedir option to config file
- remove -C option from command line

ok florian

acme-client use configuration file [1 of 5]

start using the configuration file and delete command line arguments:

-a agreement -> agreement url ...
-c certdir -> domain certificate "path"
-f accountkey -> account key "path"
-k domainkey -> domain key "path"
-s authority -> sign with "name"

new argument:
-f configfile

the changes needed to use the new configuration are local to main.c for now.
While the configuration could be passed directly to netproc(), keyproc() etc,
the diff is smaller this way.

This also removes the multidir (-m) mode for now - specify different paths in
each domain {} block instead.

ok florian

add a config file parser to acme-client (unused at the moment, so that
it can be worked on in the tree).
ok florian@ deraadt@

A small amount more KNF to make this easier on our eyes. Not going further
than this.
ok florian

hoist local variable initialization directly into the definitions,
rather than doing it right afterwards.
ok florian

Inconsistant use of case (value): idiom drives me nuts, I have no idea
what purpose it was intended to serve.

update usage(); pointed out by jmc@

Implement table driven selection to which ACME authorities we can
talk.
Suggest by and OK deraadt, OK benno.

(Later on deraadt and benno discussed if this should be handled with a
config file. This seems to be good enough for now. We can do a config
file later.)

we don't have config.h

line wrap usage

Collapse krazy abstractions for other sandbox models, using chroot and
pledge directly as needed.
ok florian

knf

whitespace

revoke -> revocate, to avoid alias
ok florian

store files in directories called acme

Import Kristaps' letskencrypt and call it acme-client in tree.
OK to get it in deraadt@ (and probably beck@)

At least deraadt@, beck@ and otto@ are fine with the name and the
disagreements stopped.

Track key type (RSA or ECDSA) in an enum and clean up a bit while here.
Originaly from Renaud Allard following input from benno, tweaked by me.
OK benno

use acme-client to sign certificated with ecdsa keys

diff from Renaud Allard <renaud@allard.it>, ok to get in from florian@

Remove A and D flag, they are superfluous.
One could always use them on the command line and acme-client would do
the right thing.

Implement RFC 8555 "Automatic Certificate Management Environment
(ACME)" to be able to talk to the v02 Let's Encrypt API.

With this acme-client(1) will no longer be able to talk to the v01
API. Users must change the api url in /etc/acme-client.conf to
https://acme-v02.api.letsencrypt.org/directory
Existing accounts (and certs of course) stay valid and after the url
change acme-client will be able to renew certs.

Tested by Renaud Allard and benno
Input & OK benno

remove free() before exit() from main(), in ongoing work to make
things more readable. otto notes that free() does some checks, but
in this case readability is better than complete cleanup.
ok florian@ deraadt@

use temporary strings for calls to basename(), because it may change
the contents of its argument (on other platforms). Also strdup() the result,
because basename returns a pointer to static memory.
ok florian@

fix dirname usage in acme-client: dirname() can modify its argument
(it does not on OpenBSD) so we need to us a copy of the string. In
addition, copy the result of dirname() as well, because it's static
storage and if we call dirname() again it will be overwritten.
Original problem noted and fix suggested by Wolf (wolf AT wolfsden DOT
cz)
ok florian@

add an extra check for existing "api ..." line in the config.
move the check of existing "account ..." line from main.c to the parser.
ok deraadt@

cleanup initialization of chngdir. inspired by Ross L Richardson.
ok tb@

replace exit() with return(), from Ross L Richardson.
ok tb@ (previous 3 commits to main.c as well)

replace warn() + exit() with err()
From Ross L Richardson.

line too long and whitespace. From Ross L Richardson.

reorder option parsing to be alphabetical/same order as usage and
manpage. From Ross L Richardson.

Deprecate agreement url config option and get the information from the
directory call. This way we don't need to update the acme-client.conf
file every time it changes. Still parse the option, ignore and warn about
it for a release. Sysmerge should be able to handle the removal.
"nice" deraadt@
OK benno

Use basename(3) instead of hand rolling it.
Gets rid of double slashes, pointed out by Raf Czlonka (rczlonka at
gmail), thanks!

OK benno

pull root check up, the parser will bomb out anyway
OK benno

Make returns consistent and remove parentheses per style(9).

Transformed with coccinelle.

Requested by and ok tb@

Complete jsing's coccinelle cleanup... by hand
ok jsing

errx() does not need extra newline

Replace comparisons between a constant or enum and an expression, with
a comparison between the expression and the constant or enum. This
significantly improves readability.

Transformed with coccinelle.

Requested by deraadt@

tweak previous;

move config data structures mostly to TAILQs, because that way we
preserve order.

ok florian@

it is ok to only specify a full chain certificate
OK benno

add option 'domain full chain certificate "path"',
revokation works, the fullchain file will be unlinked.

ok florian

Implement domain chain certificate.
OK benno

Split certificate file from config file into certdir and certfile.
This way we can still chroot to certdir but the the certificate file
is not fixed to "cert.pem".

Writing of chain.pem and fullchain.pem is currently broken with this.

OK benno

Remove backup option. This is not acme-client's business; also it gets
in the way.
OK benno

We are only dealing with one domain on the command line
OK benno

typo; ok benno

typo; ok benno

acme-client use configuration file [5 of 5]
implement new -n option to check and print configuration

ok florian

acme-client use configuration file [4 of 5]

fix getopt()

ok florian

acme-client use configuration file [3 of 5]

change command line options:

n -> A new Account key
N -> D new Domain key

With this acme-client has these main usage patterns:

* create new Account Key and Domain Key and get a certificate:

acme-client -A -D www.example.com

* renew certificate:

acme-client www.example.com

* revoke certificate:

acme-client -r www.example.com

ok florian

acme-client use configuration file [2 of 5]

- add challengedir option to config file
- remove -C option from command line

ok florian

acme-client use configuration file [1 of 5]

start using the configuration file and delete command line arguments:

-a agreement -> agreement url ...
-c certdir -> domain certificate "path"
-f accountkey -> account key "path"
-k domainkey -> domain key "path"
-s authority -> sign with "name"

new argument:
-f configfile

the changes needed to use the new configuration are local to main.c for now.
While the configuration could be passed directly to netproc(), keyproc() etc,
the diff is smaller this way.

This also removes the multidir (-m) mode for now - specify different paths in
each domain {} block instead.

ok florian

add a config file parser to acme-client (unused at the moment, so that
it can be worked on in the tree).
ok florian@ deraadt@

A small amount more KNF to make this easier on our eyes. Not going further
than this.
ok florian

hoist local variable initialization directly into the definitions,
rather than doing it right afterwards.
ok florian

Inconsistant use of case (value): idiom drives me nuts, I have no idea
what purpose it was intended to serve.

update usage(); pointed out by jmc@

Implement table driven selection to which ACME authorities we can
talk.
Suggest by and OK deraadt, OK benno.

(Later on deraadt and benno discussed if this should be handled with a
config file. This seems to be good enough for now. We can do a config
file later.)

we don't have config.h

line wrap usage

Collapse krazy abstractions for other sandbox models, using chroot and
pledge directly as needed.
ok florian

knf

whitespace

revoke -> revocate, to avoid alias
ok florian

store files in directories called acme

Import Kristaps' letskencrypt and call it acme-client in tree.
OK to get it in deraadt@ (and probably beck@)

At least deraadt@, beck@ and otto@ are fine with the name and the
disagreements stopped.

use acme-client to sign certificated with ecdsa keys

diff from Renaud Allard <renaud@allard.it>, ok to get in from florian@

Remove A and D flag, they are superfluous.
One could always use them on the command line and acme-client would do
the right thing.

Implement RFC 8555 "Automatic Certificate Management Environment
(ACME)" to be able to talk to the v02 Let's Encrypt API.

With this acme-client(1) will no longer be able to talk to the v01
API. Users must change the api url in /etc/acme-client.conf to
https://acme-v02.api.letsencrypt.org/directory
Existing accounts (and certs of course) stay valid and after the url
change acme-client will be able to renew certs.

Tested by Renaud Allard and benno
Input & OK benno

remove free() before exit() from main(), in ongoing work to make
things more readable. otto notes that free() does some checks, but
in this case readability is better than complete cleanup.
ok florian@ deraadt@

use temporary strings for calls to basename(), because it may change
the contents of its argument (on other platforms). Also strdup() the result,
because basename returns a pointer to static memory.
ok florian@

fix dirname usage in acme-client: dirname() can modify its argument
(it does not on OpenBSD) so we need to us a copy of the string. In
addition, copy the result of dirname() as well, because it's static
storage and if we call dirname() again it will be overwritten.
Original problem noted and fix suggested by Wolf (wolf AT wolfsden DOT
cz)
ok florian@

add an extra check for existing "api ..." line in the config.
move the check of existing "account ..." line from main.c to the parser.
ok deraadt@

cleanup initialization of chngdir. inspired by Ross L Richardson.
ok tb@

replace exit() with return(), from Ross L Richardson.
ok tb@ (previous 3 commits to main.c as well)