Edward Bartolo <edbarx@???> wrote:

> Keep in mind that someone from the audience may tell you that
> fine control requires the knowledge of complicated shell scripting and
> the knowledge of how diverse programs are configured in their
> configuration files.

Just like Windows, with PowerShell - and a confusingly similar but different PowerShell for Exchange - and all those ".ini' files. I try and stay out of "doing Windows" while working in a mostly Windows-centric outfit, but one thing I have learned is that even some fairly basic tasks need "complicated shell scripting" (aka, magic PowerShell incantations) to do - this is, of course, invisible to those who's skill level has never exceeded the basics that can be done with the "click things randomly till it seems to work" GUI. I've also observed that it's quite easy to screw things up (a little knowledge ...) but impossible to track down later as these incantations make invisible changes - very unlike making changes that are easy to see in a script/config file.

And the ultimate config file in the Windows world is of course ... the Registry. That small, easy to navigate, and not at all easy to damage, config database :D


Lars Noodén <lars.nooden@???> wrote:

> Even within Debian, it ended on being one single person who made the
> call to deploy systemd. Look at the ranking for the official tallies:
>
> https://lists.debian.org/debian-ctte/2014/02/msg00402.html
>
> He made the decision and it cascaded down to well over 300 derivatives:


Also, have a link ready to the analysis someone did showing that in fact "going SystemD" was a minority vote once you look at the options carefully and exclude the ones that didn't really vote for SystemD but were counted as such. Sorry, don't have the link, but I know someone posted it here not all that long ago.



Another "weapon" to have in reserve would be a list of serious bugs in stuff LP decided "had to be remade, well because". It's a tricky one to do, because you need to show that those running the project have a track record for really bad code - but without it being possible for someone to turn it around and label it as an attack on those people themselves. Eg, "you have to be careful with SystemD because LP has a history of writing crap code" can easily (but falsely) be rebutted as "so you are lowering your argument to personal attacks then ?".