Re: ChangeLog rotation (or... eliminating ChangeLogs)

[Matt Rice <ratmice at gmail dot com>]

On Tue, Jan 10, 2017 at 12:11 PM, Cary Coutant <ccoutant@gmail.com> wrote:
> What's the rationale for rotating ChangeLogs every year? To me, it's
> much simpler to have a single monolithic file -- especially
> considering that wildcard substitution puts the files in ascending
> order, while searching within a file returns results in descending
> order.
>
> Gold still has less than 20K lines total. I'm considering
> consolidating them back into a single ChangeLog (the first rotation
> for gold happened for 2016), but I don't really know what the issues
> are, and I'm willing to live with the separate files if there's a good
> reason.
>
> Even gdb has only 266K lines total in all of its ChangeLogs. Is that
> so unmanageable? Maybe we could tweak the rotation script so that it
> only rotates ChangeLogs that have grown past some threshold (say, 100K
> lines)?
>
> When Alan did the rotation last year, he suggested simply getting rid
> of them. His suggestion was met with thunderous support of one
> response (Joel). Add my vote to that! I'd be happy to have the commit
> logs serve the purpose.

While i highly doubt this would be the rationale (something more
benign like they got large and eventually it turned into a yearly
event)
But I have seen it stated before (on multiple occasions) that GNU
style ChangeLogs may result in a pathological case with git's delta
compression, I haven't tested this out, however if true it would seem
an argument in favor of rotation at least/especially for larger
ChangeLog files.

[1] https://lwn.net/Articles/701060/