1

One of the problems that comes up around releasing a piece of formerly proprietary software as open source is the necessity to "scrub the history" of potentially embarrassing or legally hazardous material, such as poorly-chosen commit messages. This can lead to project teams who have authorization to release software and publish it, but only if they delete all repository history by creating a new repo using a clean file copy -- not because anything bad is known to be in the history, but because the resources to inspect it aren't available.

My question regarding this practice is: have there been any documented occurrences where doing a "no history" release of software resulted in some kind of severe failure?

By "severe failure" I mean things like: project becomes a pariah and has few users and no contributors, or company gets sued, or lack of history leads to a hostile fork, etc. Something bad enough to make you doubt the wisdom of releasing at all.

I'm aware of the hypothetical problems, but I can't think of anywhere those have actually happened.

  • What about looking at projects that have been successful (or have they), and were released beginning with a clean file copy (e.g. svn export). Also what is the measure of success or failure? As an example consider OpenOffice.org, which was released in as source 2000, even though the history extends much farther into the past (and source or history of prior versions was never released AFAIK). OpenOffice.org was eventually forked into LibreOffice, though, so is that a success or a failure? – Brandin Jan 29 at 13:15
  • I'm talking about specific, visible failures that are attributable to the lack of history. Such as, for example, getting sued because of something that's in the undisclosed history, or getting blocked from government adoption becuase of the history, or soemthing else. – Josh Berkus Jan 29 at 18:52
  • Also, OpenOffice was released with most of its history. I know, I was working with the Sun team that had to spend 6 months "scrubbing" the history at the time. – Josh Berkus Jan 29 at 18:53
  • Gotta love the -1s on any question which isn't about licensing. – Josh Berkus Jan 30 at 19:01
  • The problem with this question is it's not well enough defined, and not answerable. Suppose we look and find a Project X that "failed" (what counts as "failed" is a bit fuzzy but you attempted to define it), and you see that that project didn't release the "history" (what exactly qualifies as 'releasing the history' is not defined at all. Nowadays we might use Git commit messages, but for a project that started 20 years ago?). So would you then say "ah hah! Project X failed because they didn't release the 'history'." No, that's only one possible reason, and probably not a likely one. – Brandin Jan 30 at 19:52
3

The oVirt project originated as closed source product that belonged to a company called Qumranet, before Qumranet was bought by Red Hat, and its closed assets were opened. oVirt's two main components, oVirt Engine and VDSM both lack history beyond this point - they both start with mega-commits that introduce fully fledged projects (which of course were extended and improved considerably since).

Neither of these projects ever created a strong developer community outside of Red Hat (although they do have a successful user community). Having said that, it's hard to determine whether this is due to the lack of history or another reason.

  • Yeah, I wouldn't attribute that to the lack of history. – Josh Berkus Jan 29 at 18:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.