[OE-core] Fetcher broken?

[Andreas Müller]

On Mon, Feb 22, 2016 at 10:35 PM, Burton, Ross <ross.burton at intel.com> wrote:
>
> On 22 February 2016 at 21:19, Andreas Müller <schnitzeltony at googlemail.com>
> wrote:
>>
>> > Did the following re-try download it with different (correct)
>> > checksum? Then it would be correct behavior if you had some
>> > e.g. incomplete flex-2.6.0.tar.bz2 tarball laying around in downloads
>> > directory.
>> >
>> I did a -ccleanall and repeated -> same behaviour
>
>
> Can you download the tarball manually and see what happens?
>
> $ wget http://downloads.sourceforge.net/flex/flex-2.6.0.tar.bz2
> --2016-02-22 21:27:27--
> http://downloads.sourceforge.net/flex/flex-2.6.0.tar.bz2
> Resolving downloads.sourceforge.net... 216.34.181.59
> Connecting to downloads.sourceforge.net|216.34.181.59|:80... connected.
> HTTP request sent, awaiting response... 302 Found
> Location:
> http://netcologne.dl.sourceforge.net/project/flex/flex-2.6.0.tar.bz2
> [following]
> --2016-02-22 21:27:28--
> http://netcologne.dl.sourceforge.net/project/flex/flex-2.6.0.tar.bz2
> Resolving netcologne.dl.sourceforge.net... 78.35.24.46, 2001:4dd0:1234:6::5f
> Connecting to netcologne.dl.sourceforge.net|78.35.24.46|:80... connected.
> HTTP request sent, awaiting response... 200 OK
> Length: 1634276 (1.6M) [application/octet-stream]
> Saving to: 'flex-2.6.0.tar.bz2'
>
> flex-2.6.0.tar.bz2                 59%[===================================>
> ] 942.21K  66.6KB/s    eta 7s     5flex-2.6.0.tar.bz2
> 59%[===================================>                         ] 954.15K
> 64.2KB/s    eta 7s     fflex-2.6.0.tar.bz2
> 60%[====================================>                        ] 972.44K
> 66.1KB/s    eta 6s
> flex-2.6.0.tar.bz2
> 100%[============================================================>]   1.56M
> 31.5KB/s    in 24s
>
> 2016-02-22 21:27:53 (65.8 KB/s) - 'flex-2.6.0.tar.bz2' saved
> [1634276/1634276]
>
> ~
> $ md5 flex-2.6.0.tar.bz2
> MD5 (flex-2.6.0.tar.bz2) = 266270f13c48ed043d95648075084d59
>
> (don't laugh at my download speed, it's not fun working on 3G)
>
> So that md5 matches what the recipe says (26627... and 24e611... for md5 and
> sha) and I've just verified that it builds fine after a cleanall.    You
> definitely need to compare what you're downloading with everyone else.
>
[superandy at localhost openembedded-core]$ wget
http://downloads.sourceforge.net/flex/flex-2.6.0.tar.bz2
--2016-02-22 22:45:44--
http://downloads.sourceforge.net/flex/flex-2.6.0.tar.bz2
Resolving downloads.sourceforge.net (downloads.sourceforge.net)... 216.34.181.59
Connecting to downloads.sourceforge.net
(downloads.sourceforge.net)|216.34.181.59|:80... connected.
HTTP request sent, awaiting response... 302 Found
Location: http://netix.dl.sourceforge.net/project/flex/flex-2.6.0.tar.bz2
[following]
--2016-02-22 22:45:45--
http://netix.dl.sourceforge.net/project/flex/flex-2.6.0.tar.bz2
Resolving netix.dl.sourceforge.net (netix.dl.sourceforge.net)... 87.121.121.2
Connecting to netix.dl.sourceforge.net
(netix.dl.sourceforge.net)|87.121.121.2|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1634276 (1.6M) [application/octet-stream]
Saving to: ‘flex-2.6.0.tar.bz2’

flex-2.6.0.tar.bz2
100%[==============================================================================================>]
  1.56M  1.55MB/s    in 1.0s

2016-02-22 22:45:47 (1.55 MB/s) - ‘flex-2.6.0.tar.bz2’ saved [1634276/1634276]

[superandy at localhost openembedded-core]$ md5sum flex-2.6.0.tar.bz2
266270f13c48ed043d95648075084d59  flex-2.6.0.tar.bz2

So download seems fine.

I also checked in oe download dir and get

[superandy at localhost download]$ md5sum flex-2.6.0.tar.bz2
266270f13c48ed043d95648075084d59  flex-2.6.0.tar.bz2

Also fine.

So I did again: -ccleanall / rebuild -> error is gone

Seems there was an automatic (re)download when I saw the first
warnings. With older sources of bitbake I have not seen such thing: In
case a checksum error was detected, the build was aborted and the new
checksums were logged to insert in the recipe. So I would now consider
this as solved and I saw a new feature in action - thanks for support.

Andreas