Closed BrianGilbert closed 6 years ago
####
# Shall we package docker installers for mac and windows with the archive?
# !!You don't need to hit enter!!.
####
I apparently do need to hit something :) Maybe Y/n prompt? What's the intent for a normal build? No docker?
Hitting "y" worked for me the first time. The second time I did it (after validating the downloaded versions) I didn't want to bother, so hit "n", and nothing happened. Despite the instructions I also hit "<n><\n>
. Turned out that one of those worked, but it gave no feedback at all so I was really wondering, Eventually it showed download results while I was writing this.
/Users/rfay/go/src/github.com/drud/quicksprint/package_additions.sh: line 9: xz: command not found
I'm doing a brew install xz
now, and very few users will ever run this script. However, it might want to test for the existence if necessary tools at the beginning instead of at the end. command -v
might do well, https://stackoverflow.com/a/677212/215713
At the end of the script it displays everything in the drupal_sprint_package staging dir, which doesn't seem very useful. It's a result of printf "${GREEN}The sprint tarballs and zipballs are in $(ls $STAGING_DIR_BASE/drupal_sprint_package*).${RESET}\n"
- How about we change that ls to something more specific so it will only show the tarball and zipball?
Also, I note that v0.0.6 seems to be wired into the QUICKSPRINT_RELEASE. It would be loads better to get it from git. Let's make QUICKSPRINT_RELEASE=$(git describe --tags --always --dirty) - that will get the current tag in there, or if it's not on a tag, something like the nice tag of this PR currently, v0.0.5-12-g70fbcaf
The sprint tarballs and zipballs are in /Users/rfay/tmp/drupal_sprint_packagev0.0.6.tar.gz
/Users/rfay/tmp/drupal_sprint_packagev0.0.6.zip
/Users/rfay/tmp/drupal_sprint_package:
COPYING
SPRINTUSER_README.md
bin
ddev_tarballs
install_ddev.cmd
install_ddev.sh
licenses
sprint.tar.xz
start_sprint.cmd
start_sprint.sh.
STAGING_DIR_NAME=drupal_sprint_package
STAGING_DIR_BASE=~/tmp
STAGING_DIR=$STAGING_DIR_BASE/$STAGING_DIR_NAME
These are all currently set up so the staging directory and the contents of the tarball are the same, so if I untar the sprint package to test it, it overwrites the staging directory (or perhaps would have things in there it shouldn't). Could we 1) Change the name of the tarball to something else? or 2) Change the name of the staging directory? Maybe both?
Let's make the xz unpack of the drupal repo quiet (in start_sprint.sh/cmd) instead of showing all that huge extraction of the .git directory and everything. That's pretty foreign to most people.
start_clean.sh doesn't start the ddev site. Shouldn't it? At the end of the the script it errors? out with Project is not currently running. Try 'ddev start'.
I think it doesn't work.
I'm not sure the "clean" idea is meaningful. What is "clean" supposed to be? Blow away everything?
I am able to ddev start
and get to the install screen. However, it doesn't have a default db loaded, probably you want them to go through the install process?
Oh but I have only tested on Mac so far. I will try windows and Linux this weekend.
@rfay So based on your last comments... pull this and then submit additional PRs after? I would rather adopt that approach before my next round of testing. I'm on a fiber connection, so I can push a release for others to start working against for next round.
@rickmanelius that will make it hard to track these comments, unless new issues are created
@BrianGilbert Fair enough. I'll work against this branch and rebuild/test locally.
Performed by Rick Manelius as of 2018-04-06
I’m using a laptop with the following high-level specifications:
go
, ddev
brew uninstall go
&& brew uninstall ddev
git clone git@github.com:drud/quicksprint.git
git fetch --all
git checkout 6c13e62
cd ~/Sites/sprint
cd sprint-20180406-1515
ddev start
.ddev start
does bring me up to an active site, I could infact ignore ./start_clean.shddev start
first and then execute to make this consistent.ddev exec drush user-login
ddev start
first and then execute to make this consistent.Open items/questions:
Testing so far, I can now get the zips generated. Moving on to testing as a macOS user. About to test as an end-user after I remove Docker for Mac and ddev. I am noting the one dependency of having docker installed to make the build, but that shouldn't trip too many people up.
Finished and updated my review for macOS. Overall, the language and guidance between steps is much better then my previous pass https://github.com/drud/quicksprint/issues/21#issuecomment-378830583. I was also successful, which is great! I believe a few people have mentioned testing Linux/Windows to confirm the ability to get to a success.
FWIW, I put a few other recommendations for either messaging (when something is happening that takes a while) or potentially re-ordering some things (such as needing to run ddev start
before running the cleanup script). That said, the most important thing is ensuring we have cross OS success, and other nice upgrades are just icing on the cake.
@rickmanelius I added a check that docker is running in to the relevant scripts in https://github.com/drud/quicksprint/pull/32/commits/3be763e59e1d7128fea986b418b30abfeec16816 & https://github.com/drud/quicksprint/pull/32/commits/91e5690c22c078cad34176622adbedaa4aada88c, which I think I committed after you had pulled to test
Testing this PR on Windows 10 Pro today... as soon as my kids will take a nap :P — I'll update here with any notes:
start_sprint.cmd
command inside my user's ~/Sites/sprint folder, and it opened a cmd window, did something super quick, but then closed the window and I didn't see any difference in Docker or anywhere else. Might it be failing silently? I can do a screencast or something to show you what I mean.
docker ps
didn't show any running containers.pause
s into the start cmd and what's really weird as that when I ran it as Administrator, it would bail out at the step where it started expanding the sprint.tar.xz
file, but when I ran it not as Administrator, it started extracting the files and seemed to succeed... very strange.start_sprint.cmd
stuff above, I opened the sprint folder it created, then ran the start_clean.cmd
file.
[][0m
at the start of lines (and [][33m
). This was in PowerShell.ddev start
again, and then noticed that it mentioned 'you must manually add the following entry to your hosts file' (though there weren't any directions in the sprintuser readme on how to do it...).C:\Windows\System32\drivers\etc\hosts
) as Administrator, adding the entry that was recommended.ddev-sprint-20180704-1506-db
, but couldn't find what the default user/pass/db are. Looking for that now.start_clean.cmd
command again to see if it would fix my problems... but it seemed to sit and wait after a few steps, and after waiting 5 or so minutes with no output, I just canceled it out and ran ddev start
again. That seemed to go through without any errors and got me back to my command prompt.ddev describe
command that can give things like database credentials, etc. So I ran that, found the defaults are db
/db
/db
with the hostname listed earlier. Drupal installed, yay!@geerlingguy something went wrong if you had that many files.. it should only be 31 files in 5 folders when extracting the drupal_sprint_package.zip file.
@BrianGilbert - I downloaded the release as instructed from https://github.com/drud/quicksprint/releases (drupal_sprint_package.zip). It was 1.34 GB and seemed to have all those files inside once expanded...
@geerlingguy There are heaps of fixes in this PR that aren't in the latest downloadable release. Also the last release was packaged by @rickmanelius and the package archives seem to have some weird issues/corruptions.. I've removed them now.
@BrianGilbert - Thanks; hopefully we can get this merged ASAP and get a 0.0.6 so the release packages are updated :)
I've completed the rest of my notes from the Windows 10 install process. If I can, I'll try doing it again sometime before I head out for DrupalCon if there's a newer release (no way I lug the Thinkpad with me all the way to Nashville :P ).
@geerlingguy I think I've addressed all of your points.. either already in this PR, or just now as addirtions to this PR.
PS: You don't need to know the DB details if using PHPMyAdmin thats packaged with ddev.
@BrianGilbert - Is the start_clean.cmd supposed to also end up installing Drupal? I needed the details because when I loaded the sprint project URL it took me to the Drupal installer.
@geerlingguy the start_clean.cmd script was missing a few steps, your feedback pointed me there so thanks.. I've added them in now..
@BrianGilbert - Awesome! I'm not leaving until Monday around noon, so if there's a new release between now and then I can download to the Windows machine based on this PR's work, I'll re-test and verify things are working more reliably :)
@geerlingguy I'm going to make one now.. hopefully I can upload the artifacts without it timing out this time.. (crappy australian broadband)
[#21] Ensure ddev is added to path, Let the user know we exited if they didn’t hit Y/y, add that install should be run via terminal on mac and Linux. [#26] Added error checking to all shell scripts. [#28] add release version to archive filename. [#30] Don’t install docker by default, improve messaging about Docker memory requirement. [#31] Add version number to Docker installers, remove spaces from filenames, make packaging of installers optional. Don’t automatically install docker on mac, make the user do it manually instead. Improve visibility of requirement to increase docker memory limit. [#30 #31] When running packaging script, delete and build it clean if staging dir already exists.