Re: [PATCH/RFC] Fix for default pager
- From
Jeff King <peff@peff.net>
- Date
- Jun 16, 2010, 06:28 UTC
- Message-ID
- <20100616062814.GA13481@sigill.intra.peff.net>
- In-Reply-To
- <gJV0lM_e77LzoiHR7moWdAApSZ7yI38lZ-w8kZwc97unWqtBc94nfg@cipher.nrlssc.navy.mil>
On Tue, Jun 15, 2010 at 11:11:35AM -0500, Brandon Casey wrote:
Show 6 quoted lines
> So, I think it's appropriate to set DEFAULT_PAGER on IRIX. There can't > be many users anyway. It's probably appropriate to set it on Solaris > too, if 'less' is not a commonly installed component on modern systems. > I wonder how surprised existing git users will be, for those on Solaris > platforms that have 'less' installed, when Solaris's crappy 'more' > becomes their pager.
I'm a little worried about that, too. On the other hand, wouldn't people who actually care about less have set PAGER already, to use it for things like "man"?
Show 7 quoted lines
> But, I also think it would be nice if git fell back to the 'cat' > behavior when it fails to spawn the pager, because the following error > is not very informative: > > casey@<a_solaris_box> # git log > sh: less: not found > Broken Pipe
The pager command is executed by the shell these days. Perhaps we should simply set DEFAULT_PAGER on these platforms to "less || more || cat", which seems to work from my simple tests. If that is too hack-ish (e.g., we really care about "does less exist", not "did it fail"), we can do a more invasive patch (or even provide a "git-pager" shell script helper to do a more thorough job).
-Peff