[fedora-arm] Fedora 15 ARMv5&7 Build Status

[Peter Robinson]

Hi Jon,

Feedback inline,

On Tue, Dec 6, 2011 at 8:27 PM, Jon Chiappetta
<jon.chiappetta at senecac.on.ca> wrote:
> I'm not sure if I can provide any new info here but here's where we are at number wise:
>
> Errors         Unbuilt        Different      Working        Built
> ---------------------------------------------------------------------------
> 1169 [+4]      3126 [-7]      27 [0]         26 [0]         6066 [+3]

What's the timeframe for the above stats. I know there's been more
than +3 in the last 24 hours.

> * We have 1169 error'd packages (that number seems to be slowly dropping tho)
> --- Is there a script out there that interfaces with Koji and groups together the most common build errors in a nice simple output?

Not that I'm aware of, most of the major failures that I've seen fall
into the following categories:
- Problematic package(s) being tagged in. nss was one, ppl another
- problems with koji infra like corrupt rpm on buildroot install,
build locking up etc
- Problems with vanilla upstream dist-f15 packages on arm not in
official gold fedora release upstream.
- Problems with packages with fix not upstream

> * We have 3126 packages still waiting to be built for the first time in Koji
> --- Some of those still have the .arm modifications on them tho in their ENVR
>
> * We have 27 packages that have been built but contain a different ENVR compared to the build in Primary Koji

Presumably your measuring this again dist-f15 tag? There are some
where there's been a fix upstream that was in git and build in koji
but not submitted as an update as it was an arm only fix. icu is one
example.

> * There are 26 builds currently active at the moment
>
> * We have built 6066 packages (which includes the imported noarchs)

Really? I have around 5921 packages in koji built for dist-f15:

$arm-koji list-tagged --latest dist-f15|wc
   5921   17764  419257

> - The numbers in brackets show the difference recorded in the numbers since the script was last run
> - For example, our unbuilt count went down by 7 pkgs but 4 of them turned out error and 3 of them built successfully

OK. Date/time of the current run and the previous run would be useful
so the stats have some form of meaning.

Some of the stats are useful, some are a little inaccurate, timeframes
would be useful to help judge :-)

Peter