read-stream-content-into-string should be smarter about element types
Although it's not strictly conforming, some implementations, in particular LispWorks, can be configured have a default string element type which is not character
, allowing for casually-written code to avoid consing enormous full-fat strings all the time. read-stream-content-into-string
uses with-output-to-string
with no element type option, and this both can cause nastiness in that case, and in implementations which do default to character
but have thinner strings will mean you always get enormous strings.
An easy fix is to specify the element type based on the element type of the stream.
I have a version with this change made, but I don't understand how to persuade gitlab to let me fork the repo and do a merge request (it started muttering at me about project limits and I just lost the will to live). I can send a patch or if someone can explain how to do the fork/merge-request thing I can do that.
[Patch now attached, this is against current master]