Re: [platform-testers] new snapshot available: sed-4.2.2.177-a348

[Jim Meyering]

On Fri, Dec 9, 2016 at 6:37 PM, Assaf Gordon <address@hidden> wrote:
>
>> On Dec 8, 2016, at 20:41, Jim Meyering <address@hidden> wrote:
>>
>> Please push your two commits.
>
> Thanks for the review.
> Pushed here:
>   doc: follow-up changes for manual improvements
>   
> http://git.savannah.gnu.org/cgit/sed.git/commit/?id=0989654f71308111e9751a9ddd9d24fb63e42fc0
>
>   maint: add missing fdl.texi to distribution
>   
> http://git.savannah.gnu.org/cgit/sed.git/commit/?id=77f32cf6c657ca9e9da36e6ce0132affc0322d36
>
>
>> On top of those, I wrote the following to stop version-controlling the
>> generated sed.texi:
>
> I like the idea, I wonder if we can take it a step further and do away with 
> the preprocessing completely ?
>
> The preprocessing step (sed-in.texi -> groupify.sed -> sed.texi) is there to 
> add '@group' commands
> inside long '@examples', so that multi-page examples will page-break 
> correctly.
>
> For example, this input:
>    @example
>    #!/usr/bin/sed -f
>
>    # Put 80 spaces in the buffer
>    1 @{
>      x
>      s/^.*$/&&&&&&&&/
>      x
>    @}
>    ...
>    @end example
>
> Will have '@group' added, resulting in:
>
>    @example
>    #!/usr/bin/sed -f
>
>    @group
>    # Put 80 spaces in the buffer
>    1 @{
>      x
>      s/^.*$/&&&&&&&&/
>      x
>    @}
>    ...
>    @end group
>    @end example
>
>
> The comment in the makefile says it is done "[...] since maintaining the 
> "@address@hidden group" manually is a burden, we do this automatically".
>
> I agree it is a minor burden, and 'groupify.sed' itself is quite 
> sophisticated.
>
> But it adds the burden for the building process.
> It adds just 84 instances of '@group' (diff attached),
> and I think that none of those multi-page SED examples have been changed in 
> several years - so very little burden in practice.
>
> The attached patch removes 'sed-in.texi' completely.
>
> What do you think?

I agree, and prefer your approach.
It is not worth retaining that complexity and duplication for so little benefit.

One nit: your change should also remove mention of doc/s-info.
While you're at it, please also include the part of my patch that
removed some duplication (two .texi files were listed twice).
Here's a proposed revision that does both of those things and includes
a slightly reworded commit message and a detailed listing of what
changed in doc/local.mk:

sed-doc-simlification.diff
Description: Text document