Re: [BUG] add_again() off-by-one error in custom format

From: René Scharfe <l.s.r@web.de>
Date: Jun 14, 2017, 17:34 UTC
Message-ID: <1d4c9a93-5061-96db-d8aa-e798948650dc@web.de>
In-Reply-To: <20170613222445.15575-1-szeder.dev@gmail.com>

Am 14.06.2017 um 00:24 schrieb SZEDER Gábor:

Show 13 quoted lines

> [sorry for double post, forgot the mailing list...]
> 
> To throw in a fourth option, this one adjusts the expansions' cached
> offsets when the magic makes it necessary.  It's not necessary for
> '%-', because it only makes a difference when the expansion is empty,
> and in that case
> 
>    - add_again() doesn't consider it cached,
>    - and even if it did, the offset of a zero-length string wouldn't
>      matter.
> 
>   pretty.c | 32 +++++++++++++++++++++-----------
>   1 file changed, 21 insertions(+), 11 deletions(-)

There are other modifiers, e.g. try the format '%h%>(20)%h' -- its output is obviously wrong (contains control characters), even with your patch. We'd have to update the offsets for each placeholder that changes the output of others. I don't think that approach scales, alas.

René