Joerg Sonnenberger

On Thu, Oct 08, 2015 at 09:55:04PM -0700, Tim Kientzle wrote: > -> The current archive_entry has a "wide" alternate that I would like > to deprecate. It was originally...

Actually, we can go one step further. It is not guaranteed that mbs -> wcs -> mbs is giving the same byte sequence, even when starting from UTF-8 and going...

So let's assume everything uses UTF-8. That still doesn't fix the problem of different systems representing Unicode differently. The wchar_t is just another facet of the same problem: there is...

wchar_t has _nothing_ to do with Unicode. Some implementation decide to make it UCS4, but that's an implementation detail and not a portable assumption. As I said, there are good...

If SHA-1 or MD5 is not provided by the system, they won't be used. The most popular user of SHA1 is certainly the "modern" ZIP encryption. We do not plan...

@letoams libarchive is not in a position to dictate such a default, because it is pretty much exclusive either a program choice or a format choice.

So, did you build libarchive against libxz? If not, that's expected and there is no fallback e.g. to an external xz binary here is either.

I haven't had time to update my buildbot instance in a while, so it is quite possible that it got broken by API changes in the last 2.5 years.

> It's not that those database can't support something like a `uint64_t`, it's just that they can't store the full range of `uint64_t` values. Sorry for the confusion here. They...

I'm very much against the restriction to C90, even a C90 with 64bit extensions. I also find many of the changes here annoying regressions, e.g. trailing comma at the end...