[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#828011: requested traces attached



Control: tags -1 = confirmed

On Fri, Jun 24, 2016 at 09:22:12PM +0200, Eric Valette wrote:
> Answer for: http://download.virtualbox.org/virtualbox/debian/dists/jessie/InRelease
> HTTP/1.1 200 OK
> Server: Apache
> ETag: "1e6763f0c4df7d4582dd89e6437276e0:1466107324"
> Last-Modified: Thu, 16 Jun 2016 16:39:18 GMT
> Accept-Ranges: bytes
> Content-Length: 7894
> Content-Type: text/plain
> Date: Fri, 24 Jun 2016 19:21:14 GMT
> Connection: keep-alive
> 
>  <- http:400%20URI%20Failure%0aURI:%20http://download.virtualbox.org/virtualbox/debian/dists/jessie/InRelease%0aMessage:%20Format%20de%20date%20inconnu%20[IP%c2%a0:%202.18.245.43%2080]%0aTransient-Failure:%20true

No sense this makes as the "Last-Modified" date the http binary is
complaining about here is a valid date and parses just fine here.


Until you realize that this is a locale thing. The French have a space
as thousand separator, so the parser (which I would actually consider
another stdlibc++6 bug) eats the entire string on the basis that spaces
aren't field but thousand separators (and 'Jun' looks very numberish for
some reason…).

The solution is simple – and I wonder why it wasn't in the second
rewrite as it was part of the first – just ensure we parse in a locale
with less ambiguous numeric formatting like all the locales apt
developers operate in, or C.UTF-8 for fairness.

As apt isn't really in the business of printing numbers setting
LC_NUMERIC=C.UTF-8 as a workaround should keep the l10n intact.


As I said the last two times already: How hard can date parsing be…


Best regards

David Kalnischkies

Attachment: signature.asc
Description: PGP signature


Reply to: