Closed Malabarba closed 8 years ago
I have seen this as well. This only happens for more recent versions of Emacs and for precompiled binaries. I haven't looked into the issue that careful as they are just warnings. But it would be nice to get fixed.
Yes. The biggest issue for me is that it floods the build reports. ☺ Specially on builds that run emacs many times, like https://travis-ci.org/clojure-emacs/cider/jobs/82025721
I can try to recompile the binary again and see if it helps. It will take some time though because I have to set up the VM, etc... all over again.
If you feel like it. :) It's not a huge deal.
Well, it's been bugging me too. I'll give it a try.
OMG I'm so smart sometimes!!! https://github.com/rejeep/evm#adding-a-new-emacs-version :smile:
:smile:
I don't know if Travis has some cache for downloads such as this. If not, then this does still not work.
I've been rebuilding Emacs 24.4, but I still get those messages. I even tried to touch all files before tar-ing it, but same result.
Looking at the tar-file, there should't be a problem (date is the same):
$ tar -vtf emacs-24.4-linux.tar.gz | grep package\.el
-rw-r--r-- 0 vagrant vagrant 87142 Sep 25 15:17 emacs-24.4-bin/share/emacs/24.4/lisp/emacs-lisp/package.el
-rw-r--r-- 0 vagrant vagrant 89405 Sep 25 15:17 emacs-24.4-bin/share/emacs/24.4/lisp/emacs-lisp/package.elc
Anyone else has a clue what's going on here?
No idea. =/
Maybe if you just remove the .el
files?
Would perhaps solve the issue. But I would like to know what the issue is... :confused:
Same here. =/
I've never written any Ruby, so I can't really follow the EVM code, but is the problem occurring during the installation? I've seen this before when, for example, rsync'ing a built emacs to somewhere else -- if the elc file goes across the wire before the el file, then it appears to be newer.
I could use `load-prefer-newer' as a fix but then I'd be running off el files rather than elc which might potentially produce some changed behaviour.
This is causing breakage in my tests here https://github.com/cask/cask/pull/349 because I need to check out output of Emacs.
Can I re-suggest Artur's idea -- just remove the *.el files. They aren't being used anyway!
Rsync has an option to preserve file attributes
I've checked this out now on the fix/evm-out-of-date-messages of http://github.com/phillord/m-buffer.
Two things I have noted. First, unless I have done it wrong, the standard instructions put the active emacs into /tmp/tmp/emacs-24.5
. And, second, removing the el files works a treat.
Compare https://travis-ci.org/phillord/m-buffer-el/builds/132091626 with https://travis-ci.org/phillord/m-buffer-el/builds/132092411.
The .travis.yml file looks like this...
language: generic
sudo: no
env:
# - EVM_EMACS=emacs-24.1-travis
# - EVM_EMACS=emacs-24.2-travis
# - EVM_EMACS=emacs-24.3-travis
# - EVM_EMACS=emacs-24.4-travis
- EVM_EMACS=emacs-24.5-travis
install:
- curl -fsSkL https://gist.github.com/rejeep/ebcd57c3af83b049833b/raw > travis.sh && source ./travis.sh
- evm install $EVM_EMACS --use --skip
## delete all the el files
## - find /tmp -name "*el"
## - find /tmp -name "*elc"
- find /tmp -name "*elc" | sed 's/.$//' | xargs rm -f
- cask
script:
- emacs --version
- make
I don't think the "remove all .elc files" is a good solution, because I've experienced differences between compiled and non compiled el-files so many times. So Emacs would not be run the way it will be for real.
@phillord Can you try and see if there's anything different with this branch: https://github.com/rejeep/evm/tree/copy-with-preserve-option
@rejeep I didn't remove the elc files, I removed the all el files for which there is an corresponding *elc file.
The *el files really are unnecessary -- quite a lot of packaged emacs' ship without them, and this seems to make sense here also. They are not being loaded, after all.
I will try the branch though.
@rejeep Tried https://github.com/rejeep/evm/tree/copy-with-preserve-option at https://travis-ci.org/phillord/m-buffer-el/builds/132135612. Appears to solve the problem, also. Either solution seems good to me: yours seem to fix the actual problem, mine produces a 50M smaller download.
Then I will go with my solution because I want the dist to be as similar as possible to what it would look like when installed IRL.
I merged my change to master, @phillord can you please verify again?
I'm not @phillord, but I've also had this issue and I can confirm this is now fixed for me.
I am @phillord, and it appears to be working on master. The errors from my cask PR are gone also.
@rejeep I think you are right, this is probably the cleaner solution.
👍
Thanks for the work, Johan!
Only took almost a year! 😄
See this build, for instance. https://travis-ci.org/clojure-emacs/cider/jobs/82025719
This is not a breaking bug, but it does make me wonder if something is going wrong. Should this be happening?