Re: seq feature: print letters

[Assaf Gordon]

Hello Pádraig,

Thanks for considering these patches.

On 10/07/2014 07:51 PM, Pádraig Brady wrote:
> Looks like there will be another 8.x release before 9.x opens.
> I intend to include this (and the sort/uniq/join field unification).
> Hopefully it wont be too long.

There's obviously no rush to include them, so if 8.24 is a quick bug-fix 
release, no need to delay it for them.

Regarding the sort/uniq/join:
I think the "sort" part is solid: the changes are minimal (mostly extracting 
the code to another file).
The uniq/join needs  review. I think they are good, but could always use a 
closer look.
Main  concern is introducing a regression: all tests pass with this patch, so I 
hope there are no regressions, but
I'm not sure the join/uniq tests cover all the bases.

The join/uniq need a NEWS entry, but with all the rebasing, I found "NEWS" to 
cause the most conflicts :)
I removed it from the patch, but the text could be something like:
===
  join accepts new options: --dictionary-order(-d), --general-numeric-sort(-g),
  --numeric-sort(-n), --reverse(-r) affecting key comparison. These modifiers
  make join more compatible with sort's --key specifications.

  uniq accepts a new option: --key (-k) to determine uniqueness of lines based
  on key specification, similar to sort's --key specifications.
===


Regarding the alphabet:
There are fours parts:
1. src/alphabet_data.c - the encoded alphabet data structure
2. scripts/encode_alphabets.pl - the script to generate the above file
3. src/alphabet.{c,h} - the decoder for the alphabet data structure
4. src/seq.c - command-line argument processing to call new functions.
These might be more 'controversial' compared to existing code in coreutils, so 
perhaps it will take time to review and accept.