[Vm-dev] [squeak-dev] new Cog VMs available

[Bert Freudenberg]

On 08.04.2013, at 10:31, Chris Muller <asqueaker at gmail.com> wrote:

> 
> I thought the "sequence number" (for lack of a better term), e.g.,
> 2714, was what made the names  unique.  If not, is it still needed
> too?
> 
> On Mon, Apr 8, 2013 at 11:36 AM, Eliot Miranda <eliot.miranda at gmail.com> wrote:
>> 
>> 
>> 
>> On Sun, Apr 7, 2013 at 11:01 PM, Tobias Pape <Das.Linux at gmx.de> wrote:
>>> 
>>> 
>>> Am 07.04.2013 um 01:30 schrieb Eliot Miranda <eliot.miranda at gmail.com>:
>>> 
>>>> On Thu, Apr 4, 2013 at 1:53 PM, Chris Muller <asqueaker at gmail.com> wrote:
>>>> 
>>>> Woo hoo!  Thanks Eliot!
>>>> 
>>>> Quick question, I see you're using a new naming convention, what does
>>>> 13.13 mean?
>>>> 
>>>> It means the 13th week of 2013.  So all the archives are named archive-WEEK.YEAR.SVNID.ext.
>>>> 
>>>> Hope that's not too inconvenient.  But they need unique names.  I stupidly overwrite a set of old files doing some reorganization a few weeks back and that wouldn't have happened with unique names...
>>>> 
>>> 
>>> I think this is pretty useful, however, what about
>>> 
>>>        archive-YEAR.WEEK.ID.ext
>>> 
>>> ? That way, the vms are ordered successively :)
>>> (as the current one is 13.13.… changing shouldn't be a problem ;))
>> 
>> 
>> Oops.  It is indeed YEAR.WEEK.ID.  I misspoke.  Here's the command that delivers the tag:
>> 
>> REV=`grep 'SvnRawRevisionString.*Rev:' platforms/Cross/vm/sqSCCSVersion.h \
>>        | sed 's/^.*Rev: \([0-9][0-9]*\) $";/\1/'`
>> TAG=`date +%g.%U.`$REV

Doesn't make sense to me. Builds should be reproducible, so why encode the build date in the file name? If you really need it, then the SVN version should take precedence over the build date. So "id-date" would be better, grouping same sources but different build dates together.

- Bert -