On 29 May 2015 at 12:51, Peter Eisentraut <peter_e@gmx.net> wrote:
On 5/12/15 4:33 AM, David Rowley wrote: > Shortly after I sent the previous patch I did a few more searches and > also found some more things that are not quite right. > Most of these are to use the binary append method when the length of the > string is already known.
For these cases it might be better to invent additional functions such as stringinfo_to_text() and appendStringInfoStringInfo() instead of repeating the pattern of referring to data and length separately.
You're probably right. It would be nicer to see some sort of wrapper functions that cleaned these up a bit.
I really think that's something for another patch though, this patch just intends to put things the way they're meant to be in the least invasive way possible. What you're proposing is changing the way it's meant to work, which will cause much more code churn.