[Toybox] [PATCH] patch: support GNU-style [FILE [PATCH]] arguments.

[enh]

hmm, i might have run into the -R issue by accident while testing my
change the other day. this is the most recently added item on my list
of "things to investigate later" (the patch below is just a stupid
change i made for testing)...

/tmp/toybox$ cat patch
diff --git a/lib/lib.c b/lib/lib.c
index 2250caf4..381173ee 100644
--- a/lib/lib.c
+++ b/lib/lib.c
@@ -1,5 +1,4 @@
 /* lib.c - various reusable stuff.
- *
  * Copyright 2006 Rob Landley <rob at landley.net>
  */
 /tmp/toybox$ ./toybox patch -p1 < patch
patching lib/lib.c
/tmp/toybox$ git diff
diff --git a/lib/lib.c b/lib/lib.c
index 2250caf4..381173ee 100644
--- a/lib/lib.c
+++ b/lib/lib.c
@@ -1,5 +1,4 @@
 /* lib.c - various reusable stuff.
- *
  * Copyright 2006 Rob Landley <rob at landley.net>
  */

/tmp/toybox$ ./toybox patch -R -p1 < patch
patching lib/lib.c
Hunk 1 FAILED 1/1.
 /* lib.c - various reusable stuff.
- *
  * Copyright 2006 Rob Landley <rob at landley.net>
  */

/tmp/toybox$


yeah, looks like it. if i switch instead to a first hunk that adds
something (like below) -R works.

/tmp/toybox$ cat patch
diff --git a/lib/lib.c b/lib/lib.c
index 2250caf4..b57d7970 100644
--- a/lib/lib.c
+++ b/lib/lib.c
@@ -1,6 +1,7 @@
 /* lib.c - various reusable stuff.
  *
  * Copyright 2006 Rob Landley <rob at landley.net>
+ *
  */

 #define SYSLOG_NAMES
/tmp/toybox$

On Sat, Dec 21, 2019 at 12:00 AM Rob Landley <rob at landley.net> wrote:
>
> On 12/20/19 12:48 PM, enh via Toybox wrote:
> > POSIX only mentions -i/stdin, but GNU patch -- and Larry Wall's patch 1.3,
> > found via https://en.wikipedia.org/wiki/Patch_(Unix) -- also support
> > supplying the name of the file to patch and the name of the patch file
> > as optional arguments.
>
> If Larry Wall's original patch supports it, it's not gnu style is it? :)
>
> Applied, but I noticed there's no patch.tests so started doing some simple ones,
> and I hit the corner case that each patch has TWO filenames and the debian patch
> will use _either_ filename if the first doesn't exist, and mine won't.
>
> The thing is... nobody's ever _noticed_. I can add this fallback behavior,
> but... I think the current code is using the +++ line when the first thing we do
> is + and the --- line when the first thing we do is - and that seems to be more
> correct? (Modulo -R which again, does not appear to have come up?)
>
> I'm... not sure if I _should_ add the extra plumbing? (I'm pretty sure I was
> aware of this issue way back when and chose _not_ to, although I'd have to dig
> into the mailing list or blog to see if I documented _why_. There's a lot of
> https://landley.net/notes-2007.html#29-12-2007 entries from back then. And yes
> https://web.archive.org/web/20060325013934/http://kernelslacker.livejournal.com/7336.html
> is why I didn't implement fuzz factor support...)
>
> Rob