Promoting the use of Mercurial; was: Re: gnome dvcs survey results

[Peter Arrenbrecht]

On Sat, Jan 10, 2009 at 6:52 PM, Giorgos Keramidas
<keramida at ceid.upatras.gr> wrote:
> On Fri, 09 Jan 2009 23:09:27 -0600, Matt Mackall <mpm at selenic.com> wrote:
>>On Fri, 2009-01-09 at 23:43 -0500, Theodore Tso wrote:
>>> This assumes that tag _are_ history. Another way to think about
>>> things is to say that tags are a *pointer* to history. Some
>>> historiams claim September 4, 476 as the fall of the Roman Empire;
>>> others use 378 A.D., still others 410 A.D., and others go as late as
>>> 565 A.D.  *History* is immutable; what tags we place on history isn't
>>> necessarily part of history.
>>
>> Sure, tags are just pointers. But *tagging* is an action that is
>> history-relevant, precisely because you want to ask questions like
>> "who created or changed this pointer, when and why?"
>
> Precisely.  The tagging operation *is* history.
>
> I have worked in jobs where 'sliding a tag' was needed some times.  In
> some of these cases I used "hacks" to keep a side-history of who slided
> a tag, when it was done and why.  With 'hg tag' this isn't necessary
> because every tagging operation carries all this extra information in
> its changelog :)

To compound the matter even further, assume you have the same (sliding
or not) tag in multiple branches. For instance, version 1.1 for this
customer, and version 1.1 for that one, assuming a company produces
slightly modified customer-specific versions of a core product as
named branches. (I'm not saying anything in this scheme is good
practice, but it illustrates why one might want tag definitions to be
able to diverge, much like the rest of the DAG.)

-parren