On Wed, 19 Jul 2017 17:30:53 -, Michael Witten wrote:
> On Thu, 20 Jul 2017 00:31:25 +0800, Matt Johnston wrote:
>
>> Thanks for your patches. The Makefile dependency ones look
>> useful, I'll try and double-check them and get them merged
>> in the next week or so. P
ar as part of a larger, automated building
process should receive such a notification if the `stderr' output is
being logged for later review.
In short, this is potentially a great opportunity to clean up the source
code, establishing a more solid foundation for future development.
Sincerely,
Michael Witten
Date: Tue, 18 Jul 2017 05:02:16 -
Options are now categorized explicitly into 2 groups:
* Customizable options
These are the object-like macros in `default_options.h.in'; they are
given the prefix `CUSTOM_', and they can be overriden by the definitions
Date: Mon, 17 Jul 2017 21:11:17 -
* These macros are not used anywhere, and have therefore been removed:
MIN_DSS_KEYLEN
ENV_SIZE
* According to `CHANGES', the macro `DROPBEAR_CLEANUP' is supposed
to be disabled by default; this commit disables it by default.
* A number of the
This is a request for comments on the following patch series, which
is an attempt to establish a more consistent way of organizing how
options are defined, used, and documented; it's mainly a bunch of
refactoring:
[1] options: Give the header files names that are more consistent
[2] options:
On Sat, 15 Jul 2017 20:46:02 -, Michael Witten wrote:
> Now the comments reflect the way the code works; no code
> has been changed.
>
> In the process of updating the comments, they have also been
> copy edited; notably, old-style C comments (/**/) have been
> replaced w
Now the comments reflect the way the code works; no code
has been changed.
In the process of updating the comments, they have also been
copy edited; notably, old-style C comments (/**/) have been
replaced with new-style, one-line C++ comments (//).
For the record, `packet.c' is another file that
+-
svr-tcpfwd.c | 2 +-
sysoptions.h | 6 +-
12 files changed, 303 insertions(+), 220 deletions(-)
Sincerely,
Michael Witten
On Tue, 11 Jul 2017 05:08:56 -, Michael Witten wrote:
> Thus, for now, the minimal knowledge of prerequisites to which `make'
> has access during the initial run is basically just the one header
> which potentially needs to be made before almost everything else:
>
> de
On Wed, 12 Jul 2017 17:34:49 -, Michael Witten wrote:
> On Tue, 11 Jul 2017 05:08:56 -0000, Michael Witten wrote:
>
>> diff --git a/libtommath/Makefile.in b/libtommath/Makefile.in
>> index dbcd2a0..c06187a 100644
>> --- a/libtommath/Makefile.in
>> +++ b/libt
of these patches in a more coherent format.
Sincerely,
Michael Witten
To apply this patch, save this email to a file:
/path/to/email
and then apply it:
git am --scissors /path/to/email
Here is the scissors-line across which git will cut in order to
throw away the text above it.
8<--
It was safe to remove `default_options.h', but in my excitement to do so,
I forgot to take care of some administrative issues, namely that `git'
should be told to ignore that file, and the command `make clean' should
remove it, and there's now no need for a comment in `Makefile.in' about
the
On Tue, 11 Jul 2017 05:08:56 -, Michael Witten wrote:
> diff --git a/libtommath/Makefile.in b/libtommath/Makefile.in
> index dbcd2a0..c06187a 100644
> --- a/libtommath/Makefile.in
> +++ b/libtommath/Makefile.in
> @@ -17,7 +17,7 @@ endif
> ifneq ($V,1)
> @echo &qu
Whenever building `libtommath', `make' would complain to me that
it could not find the `arch' command.
According to `info arch':
`arch' prints the machine hardware name, and is equivalent to
`uname -m'... `arch' is not installed by default, so portable
scripts should not rely on its
The file `libtommath/makefile.include', which is supplied by the
libtommath sources, already supplies its own `clean' target; besides
the fact that GNU `make' frowns upon multiple recipes for the same
target, it is likely the case that libtommath's own `clean' target
is worth using.
Indeed, the
Now, it will only be necessary to maintain `default_options.h.in',
though it does seem like a lot of trouble just to avoid having
users deal with #ifndef/#endif statements.
---
default_options.h | 466 --
1 file changed, 466 deletions(-)
delete
This commit provides 3 useful results:
* It allows parallel make (`make -j') to function correctly.
* It allows `make' to be run from within the libtom(crypt|math)
directory without having to worry about the state of
prerequisites provided by Dropbear or by the other libtom*.
* It
them as follows:
git am /path/to/patches/*
Sincerely,
Michael Witten
These targets have been removed:
multibinary
multilink
link%
The target `multilink' isn't explicitly used anywhere, and it doesn't
seem like it would be that useful for most users, and can thus be
removed.
The targets `multibinary' and `link%' are only referenced by the
target
This makefile is sprawling enough that there is more value in
placing `.PHONY' explicitly next to each phony target, rather
than trying to group all the phony targets together as
prerequisites to `.PHONY' target. In my experience, it's much
easier to know what's going on by being able to look at
Due to the fact that platforms disagree about which file-name
extension an executable should have, subtle problems have accrued;
notably, it became the case that build targets were no longer
explicitly being treated as actual, real files, but sometimes as
implicitly phony targets (that is, a
improvements,
such as getting parallel make (`make -j') to handle
dependencies better.
Sincerely,
Michael Witten
In-Reply-To<20151229150334.gb27...@ucc.gu.uwa.edu.au>
References: <cover.1449517677.git.mfwit...@gmail.com>
<20151229150334.gb27...@ucc.gu.uwa.edu.au>
On Tue, 29 Dec 2015 23:03:34 +0800, Matt Johnston wrote:
> On Mon, Dec 07, 2015 at 10:42:59PM -, Michael
On Tue, 29 Dec 2015 23:03:34 +0800, Matt Johnston wrote:
> On Mon, Dec 07, 2015 at 10:42:59PM -0000, Michael Witten wrote:
>> User names have hitherto been handled neither consistently nor well; this
>> series alleviates at least some of the issues.
>>
>> Fea
On Mon, 11 Jan 2016 16:37:09 -, Michael Witten wrote:
> Message-ID: <2666123da20b47a589b819f37bd4cc54-mfwit...@gmail.com>
> In-Reply-To<20151229150334.gb27...@ucc.gu.uwa.edu.au>
> References: <cover.1449517677.git.mfwit...@gmail.com>
> <2015122
Date: Wed, 4 Nov 2015 05:56:09 -
---
scp.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/scp.c b/scp.c
index 70f45e3..b9aad3f 100644
--- a/scp.c
+++ b/scp.c
@@ -188,7 +188,7 @@ do_cmd(char *host, char *remuser, char *cmd, int *fdin, int
*fdout, int argc)
if
Date: Wed, 4 Nov 2015 20:33:19 -
It would seem that it's standard practice not to include a newline in the
message
text, but that results in poor formatting, as a shell's command line then begins
on the line of the error message itself.
This commit simply instructs `fatal()' to append a
Date: Wed, 4 Nov 2015 06:30:28 -
The program `dbclient' understands the option `-v' only when
it is built with the macro DEBUG_TRACE defined; ergo, the
program `scp' should pass `-v' to `dbclient' only when the
macro DEBUG_TRACE is defined.
---
scp.c | 4
1 file changed, 4 insertions(+)
Date: Wed, 4 Nov 2015 05:37:08 -
The variables `*suser' and `*tuser' should be of type `const char' rather
than just `char'. Furthermore, the function `okname' should be declared
as being `static'.
While these changes could be applied to other identifiers, they are valuable
in this commit as
Date: Wed, 4 Nov 2015 05:41:53 -
There's no reason to assume that there is a user name; userspace might be so
primitive that there is no such thing, and `scp' should continue to run as
long as the user explicitly specifies a user name on the command line.
So, this commit introduces a new
Date: Fri, 6 Nov 2015 20:05:54 -
---
scp.c | 7 ++-
1 file changed, 2 insertions(+), 5 deletions(-)
diff --git a/scp.c b/scp.c
index dd92451..9a24490 100644
--- a/scp.c
+++ b/scp.c
@@ -1161,11 +1161,8 @@ get_user_name()
static const char *user_name = NULL;
if (user_name
Date: Fri, 6 Nov 2015 19:12:49 -
It seems unnecessarily draconian to treat an unspecified user name as a fatal
error;
indeed, `dbclient' attempts to use "unknown" in this case, which is better than
nothing, and similarly warns the user about the problem via the usual machinery.
Furthermore,
Date: Fri, 6 Nov 2015 23:13:25 -
The variables with the name `remuser' cannot hold the value `NULL'; this commit
removes the code that handles that case.
---
scp.c | 16 ++--
1 file changed, 6 insertions(+), 10 deletions(-)
diff --git a/scp.c b/scp.c
index a19d28f..0adb590 100644
Date: Sat, 7 Nov 2015 03:27:36 -
The program `dbclient' can already handle parsing such arguments, including
determining
the current user's name.
It should be noted that relying on this centralized machinery changes the
behavior:
Whereas before, `@host' (i.e., the empty string as the user
Date: Sat, 7 Nov 2015 03:48:23 -
It seems that it has no purpose.
---
scp.c | 1 -
scpmisc.c | 20
scpmisc.h | 1 -
3 files changed, 22 deletions(-)
diff --git a/scp.c b/scp.c
index 9bf22e1..599042c 100644
--- a/scp.c
+++ b/scp.c
@@ -174,7 +174,6 @@
Date: Sat, 7 Nov 2015 16:55:53 -
This is in preparation for treating `own_user' with less explicit safety.
---
runopts.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/runopts.h b/runopts.h
index 7d6ae06..0a97937 100644
--- a/runopts.h
+++ b/runopts.h
@@ -127,7 +127,7 @@
Date: Sat, 7 Nov 2015 17:13:01 -
The variable `*cli_opts.own_user is of type `const char', and it doesn't seem
to be freed anywhere.
---
cli-runopts.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/cli-runopts.c b/cli-runopts.c
index e8cb313..c4cd12c 100644
---
37 matches
Mail list logo