[sleuthkit-developers] versioning

[Stuart M. <st...@ap...>]

On 08/13/2015 10:13 AM,
sle...@li... wrote:
>
> Message: 2
> Date: Tue, 28 Jul 2015 09:41:15 -0400
> From: Brian Carrier <ca...@sl...>
> Subject: Re: [sleuthkit-developers] whither 4.2?
> To: Jon Stewart <JStewart@StrozFriedberg.com>
> Cc: "sle...@li..."
> 	<sle...@li...>
> Message-ID: <A21...@sl...>
> Content-Type: text/plain; charset=utf-8
>
> Hi Jon,
>
> Yea, I?m sorry about that. Whenever we do an Autopsy release, I always mean to do a TSK release too, but it falls off the list. 
>
> I?ll try to get a 4.2 release out this week.  In the early Fall, there will be new big releases of Autopsy and TSK that add collaboration features.  For TSK, that means support for PostgreSQL databases in addition to SQLite.  We?ll have to decide then if that means it is 4.3 or 4.2.1. 
>
> brian
>


Hi Brian, all,

I don't want to make trouble for you in versioning, but have you
considered following the idiom of semantic versioning
(http://semver.org)?  Though TSK has command line tools, it is also a
library (and thus an API) that users, myself included, write code
against.  Semver suggests that if a new release of the API breaks
compatibility with respect to existing code, its major number needs
changing etc.

What I like about semver is that if I know you are adhering to it, and I
see a Sleuthkit version 5.0 out, and my apps were coded against 4.3, I
KNOW to be wary of trying to link against 5.0, since I EXPECT api
incompatibilities.  I also EXPECT that linking against say 4.4 should be
fine for me, since all the functionality of 4.3 is preserved and that I
might want to look at the new features added to 4.4 from 4.3.

This all sounds very easy in theory and I can fully appreciated it's
never that easy in practice, but I thought it worth mentioning semver in
this arena.

Regards

Stuart

PS My own Java bindings for TSK, which have taken me 2 years to get
github-ready, are about to be put up on github.