From: Julian Phillips <julian@quantumfyre.co.uk>
Date: Wed, 14 Apr 2010 21:28:56 GMT
Subject: Re: [RFC/PATCH v2 0/4] A new library for plumbing output
Message-ID: <0e50c39792144f5cb18260009384b4d0@212.159.54.234>
In-Reply-To: <201004142316.07947.jnareb@gmail.com>

On Wed, 14 Apr 2010 23:16:06 +0200, Jakub Narebski <jnareb@gmail.com>
wrote:
> On Wed,  14 Apr 2010, Julian Phillips wrote:
>> On Wed, 14 Apr 2010 21:10:35 +0200, Jakub Narebski <jnareb@gmail.com>
>> wrote:
>> Anyway, I have already converted blame to use the library for both
>> --porcelain and --incremental output, so it'll be in the next version
of
>> the patch series.  So you can try before you buy ...
> 
> Nice.
> 
> How did you managed to work with a bit non-standard rules of --porcelain
> format, namely maybe-quoting of filenames, and that not all lines
conform
> to "<header> SP <value> LF" syntax: group definition begins with SHA-1,
> and contents is indented with TAB?

I had to extend the output API for quoting rules.  There are now
qstr/qstrf output functions that call different backend functions.  The
normal output then applies git quoting rules to the q versions only,
whereas the JSON output treats them both the same.

For the rest of it, only the actual data is going through the output_str
etc functions the rest is using output_token that is ignored by the JSON
backend.  I actually started converting blame twice - the first time I
added an API function that allowed telling the normal output how to format
values, but I decided that I wanted to apply grouping to the structured
output, so that e.g. the author and committer information were two objects
with name, mail, date and tz members so I never finished the first attempt.

-- 
Julian