[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re:
From: |
Andreas Grünbacher |
Subject: |
Re: |
Date: |
Sun, 12 Jan 2020 05:04:07 +0100 |
Hi Vladimir,
Am Sa., 11. Jan. 2020 um 16:19 Uhr schrieb Vladimir D. Seleznev
<address@hidden>:
> The following patches implement strip value guessing option for GNU Patch.
> Please take a look at them.
>
> The idea is simple: it looks for the longest existent pathname to patch, and
> if it finds one, then it tries to patch the file.
>
> If the patches are deemed fine to be accepted, I'm ready to proceed with the
> copyright assignment.
In general, this rather seems like an option not to use. But let's see
if the idea can be salvaged.
When the first patch in the input creates a file in a sub-directory
(or removes a file and is applied in reverse), the guess will fail if
that sub-directory doesn't exist:
--- /dev/null
+++ a/dir/foo
@@ -0,0 +1 @@
+foo
I guess it would make sense for patch to refuse guessing in that case?
What should happen when patch finds multiple of the names in the patch
(pch_name)? Should the minimum strip level be used?
How does this interact with -R guessing as with the following patch
(apply in an empty directory with -p0)?
--- /dev/null
+++ bar
@@ -1 +0,0 @@
-bar
The "/dev/null" patch doesn't make sense to me.
If we want this kind of guessing, we should document how it works and
when it doesn't.
This needs to become a new long option; we can't cram it into -p.
Thanks,
Andreas