On Jun 29, 2010, at 5:57 PM, Dr. David Kirkby wrote:
On 06/30/10 01:38 AM, Mike Hansen wrote:
On Tue, Jun 29, 2010 at 5:31 PM, Dr. David Kirkby
<david.kir...@onetel.net> wrote:
I can't myself see what we gain having the diff and the modified
version.
Even if we stick to using 'cp' (which is a bad idea in my
opinion), I would
much rather create the diff myself if I want to see it.
The modified version is for copying over. The diff is stored in the
repository so that you don't need to go download old versions of
source code if you want to see what the changes are.
But if there are two three files
src/foo.c
patches/foo.c
patches/foo.c.diff
what does patches/foo.c.diff possibly give me that could could get
by running
diff src/foo.c patches/foo.c ?
Having all three files is certainly an issue, as they're unlikely to
all be kept in sync.
Experience tells me that if there is a file called 'diff' it is
quite likely
an old one which someone has not updated. So I personally never
trust their
contents - I'd just rather create it myself if I need it. Since
they can't
be trust, I don't believe they serve a useful function myself.
The spkg shouldn't get a positive review if those are not up to date.
I don't disbelieve you. But in practice I know that if I want to see
the difference between two files, I run diff myself.
Personally I believe if we added the small GNU patch utility to
Sage, it
save more space than it uses. We could over time delete a lot of
large
files, which have only small changes from original large files.
I personally thing it'd be better to just include patch.
--Mike
Me too. But I think William was quite against using 'patch'.
If a user has gcc, they almost certainly have patch (or, as mentioned,
we could provide it), so I don't think dependancies are that big of an
issue. Personally, I would rather have the patch files (as what was
changed seems to be the most important piece of data here, and we
don't have issues like we had with the recent pari spkg where the
copied files weren't updated when the sources were--patches aren't as
brittle in this way). Also, patch files contain metadata (e.g. what
files their patching, no matter how deep down the tree, and patch
programs should ignore the "header" before the patch starts where
additional useful explanations can be put. Then they could even be
applied automatically as part of the spkg installing process, rather
than having to have a list of cp commands.
Is it worth revisiting the issue?
- Robert
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
