[98388] trunk/dports/databases

[Bradley Giesbrecht]

On Oct 5, 2012, at 9:27 AM, Craig Treleaven wrote:

> At 9:03 AM -0700 10/5/12, Blair Zajac wrote:
>> On Oct 5, 2012, at 5:59 AM, Craig Treleaven <ctreleaven at cogeco.ca> wrote:
>> 
>>> At 12:02 PM +0200 10/5/12, MK-MacPorts at techno.ms wrote:
>>>> > Failing that, yes, you'd need to continue to specify a changeset. I think you can do it this way:
>>>> > github.setup        darkrose csvdb afad8eca960af3b61b0a8ee3e1c3e0db4cc5c8f5
>>>> > version             0.5.1
>>>> Yeah, just a minute ago I figured that.
>>> 
>>> Note that you don't need the full Git hash.  I believe that the first 7 characters is sufficient but I use 8.  (Eg "afad8eca".) Makes paths a little less unwieldy!
>> 
>> Given we're moving to ever longer checksums in our Portfiles, it seems best to use the full sha.
>> 
>> BTW, what happens if there is a duplicate truncated sha?
> 
> I can't find an 'official' reference for it, but I believe GitHub says that 7 characters are unique (7**26).  Anyway, if it were not unique, the MacPorts checksums would alert the user.  Seems safe to me.
> 
> I also use the short hash as part of the full version string for the port since it identifies the last commit included from the branch. This may or may not matter to other ports.

I believe there have been discussions for reasons to not use changeset hashes in MacPorts version strings.

Maybe the hashes are not always greater strings to MacPorts vercmp?


Regards,
Bradley Giesbrecht (pixilla)

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2763 bytes
Desc: not available
URL: <http://lists.macosforge.org/pipermail/macports-dev/attachments/20121005/ac2401f6/attachment-0001.p7s>