Skip to content

PSA: Spacemacs freezing? Check your savehist file

  • by

Mine was 1.6GB, beat that!

Today I was working on writing up some videos for YouTube in Spacemacs, which I’ve talked about using before, when it began periodically freezing: it wouldn’t draw anything in the window, input would buffer for several seconds at a time. Quite inconvenient. It was working fine yesterday — sorta, I’ve had one or two segfaults on loading org files — and usually works without a hitch. I glanced at top and noticed it was taking 100% CPU.

Branch master deprecated

I noticed in the recent scrollback of Spacemacs chat that someone else was having trouble with org files, so thought I’d check my own org version

and I also found out that Spacemacs itself was well behind where development is, accidentally pulled† from the master branch, which is unmaintained:


                        VERSION 0.200.14 (deprecated)


This is version 0.200.14 of the currently unmaintained master branch. The master
branch is using curated version of packages for increased stability.

However this approach proofed to be very work intensive and we have therefore
stopped supplying those versions for now.

As these packages are not longer updated you should consider migrating to the
maintained "develop" branch.

I guess that’s why I’m a bit behind?

Updating Spacemacs

There’s a couple of parts to updating components of Spacemacs. The easier of the two is to update packages– there is a link on the *Spacemacs* buffer (shown at startup by default), or you can call configuration-layer/update-packages (<leader> f e U)

To update Spacemacs itself there are instructions listed on the README, but there’s also the update section of the manual. If, like me (see † above), you end up reading both and end up with your window open on the bit for updating from master… don’t do that. Just pull from develop and rebase.

Potential org-mode issue

There is/was an issue with updating org-mode to 0.6.1; if it affects you, check the linked GitHub issue for workarounds. There is more info on the issue by jcs over at Irreal.

Savehist ballooning

Almost by happenstance I noticed that my savehist was rather large, at over a gig and a half. For a simple text file of minibuffer history, this seemed kinda large. Since backups have been on my mind in the last week (or so) I figured I’d check the file:

Yes, 1,648,980,416 bytes (1.64 GB) versus 80,315 bytes (80 kB) is a… significant difference.

Searching brings up an emacs QA with a related GitHub issue. I deleted my savehist, and I am considering using borg delete ± borg recreate to purge it from backups once I’ve checked it to see if there’s any info that would be useful for the GH issue.

It seems there’s PNG data in the kill-ring somehow:

Not sure how that got there!

Tell us what's on your mind