Add fancy install script, closes #217 #218

arichiardi · 2017-02-18T00:09:47Z

Here it is, can of course be improved on fanciness 🚬

arichiardi · 2017-02-18T00:16:58Z

Still not working, but close 😉

arichiardi · 2017-02-18T00:23:29Z

Works now! Let me know or push a commit if there are issues ok?

scottchiefbaker · 2017-02-28T21:38:04Z

I like the idea of this, but I think it needs a couple of tweaks

Some sort of error if the user doesn't have git installed
Error out if ~/bin/ isn't in the path
1. Offer an option to install to another directory in path?

Overall I'm impressed though. Good work!

arichiardi · 2017-03-01T06:03:58Z

Oh oh well, I guess I gave for granted that they would have git 😀

What the error text could be?

About the bin folder, this is actually an argument probably...

OJFord · 2017-03-01T15:43:59Z

Some sort of error if the user doesn't have git installed

Is that really necessary?

The install process doesn't depend on git, #193 is adding Mercurial support, and we're even discussing in #220 how to use dsf with non-VCS diffs.

Even if I am using dsf with git, if I have my own install script which does ... && install_dsf && install_git && ..., it's annoying if I have to change that after discovering it errors because dsf needs git at runtime (though it actually doesn't, per above) if it doesn't at install time.

arichiardi · 2017-03-01T15:47:04Z

Ah good to know,.

Also, where do folks usually install bin? I personally use $HOME/bin, but usr/local/bin is probably the classic Linux default.

OJFord · 2017-03-01T16:10:01Z

Also, where do folks usually install bin? I personally use $HOME/bin, but usr/local/bin is probably the classic Linux default.

Personally I think /usr/local/bin is better, pretty likely to already be on $PATH whereas $HOME/bin will be only for you, and others that deliberately use it. I do too, but only for small 'n-off' (for n slightly larger than 1) stuff that I've written for myself - so I'd be slightly annoyed if an install script put something third-party there. But that might just be me.

I certainly don't think ~/bin not already on $PATH should be an error. That will hit the vast majority of users, and they'd be surprised.

It would also mean this script couldn't be used in the Homebrew formula - which should IMO be a goal, (not that what I think matters - up to @stevemao 🙂) at the moment it adds a node/npm dependency, which is not necessary but presumably done for ease of maintanence, which this script can hopefully solve. It'd be nice not to have brew install all of node just for a relatively tiny bash/perl script.

arichiardi · 2017-03-01T17:38:20Z

@OJFord so can you confirm that brew installs in /usr/local/bin? Or shall I pass a parameter there? Not very familiar with brew...

scottchiefbaker · 2017-03-01T18:22:21Z

I forgot about mercurial/vanilla diff support. You can ignore my comment a missing git error message.

My vote for installation location is ~/bin/ if it's available/created. By default a lot of installs won't have this, A fall back to /usr/local/bin/, would be my second choice, but that will require root to write to.

scottchiefbaker · 2017-03-01T18:27:46Z

Think the installer should attempt to put the files in ~/bin/ IF it's already in the path. If it's not in the path attempt to put the files in /usr/local/bin but error out if the user is not root. That should cover both our install scenarios.

Perhaps this is a separate issue, but I HATE the default colors. After install people will expect the output to be the same as the screenshot in the README. Perhaps we set the colors for the user? Or offer an option to use the d-s-f "suggested colors"?

arichiardi · 2017-03-01T21:25:17Z

If we have a way to identify if we are being executed in `brew` this can be conditional as well. -- Sent from my Android device with K-9 Mail. Please excuse my brevity.

OJFord · 2017-03-01T22:29:43Z

can you confirm that brew installs in /usr/local/bin

I can.

Perhaps we set the colors for the user?

Please don't do this by default. Installing dsf shouldn't alter my non-dsf config options unless I explicitly opt in; colours are diff (git core) and diff-highlight options.

scottchiefbaker · 2017-03-01T23:32:37Z

Please don't do this by default. Installing dsf shouldn't alter my non-dsf config options unless I explicitly opt in; colours are diff (git core) and diff-highlight options.

We should definitely make it an option... "Use default d-s-f colors (will update your gitconfig)?" or something like that.

scottchiefbaker · 2017-03-06T17:16:36Z

@arichiardi where are we at on this? I'm ready to merge this if we can include an option to set the default colors.

[color]
	ui = always
[color "diff"]
	meta = yellow bold
	commit = green bold
	frag = magenta bold
	old = red bold
	new = green bold
	whitespace = red reverse
[color "diff-highlight"]
	oldNormal = red bold
	oldHighlight = "red bold 52"
	newNormal = "green bold"
        newHighlight = "green bold 22"

arichiardi · 2017-03-06T17:19:39Z

@scottchiefbaker oh cool!
I have had a pretty busy open source month 😄 Feel free to add this in and a subsequent commit can add default colors.

scottchiefbaker · 2017-03-24T15:00:49Z

@arichiardi we're really close to a 1.0.0 release of d-s-f and I really want this to be part of that. Will you have any time coming up that we can refresh this with a couple of tweaks and then I can land it?

arichiardi · 2017-03-24T16:20:09Z

Yes, I can work on it, let me know when you think changes are stable for me to start tweaking :)

scottchiefbaker · 2017-03-24T17:22:33Z

master is stable right now, so you can make changes as you see fit. As the last part of the install script I'd like to offer to set the default colors for the user. There is a new option:

diif-so-fancy --colors

That will output the commands needed to set the default colors. Perhaps we can include that in some fashion.

scottchiefbaker · 2017-03-24T21:05:51Z

As of now the only files that are needed are /diff-so-fancy and /third_party/diff-highlight/diff-highlight. Putting those in ~/bin should be sufficient.

The libexec directory is no longer needed.

scottchiefbaker · 2017-03-24T21:12:46Z

Actually for simplicity you could put both diff-so-fancy and diff-highlight in ~/bin (or anywhere in the path) and they will work just fine.

No need for a /third_party directory at all.

arichiardi · 2017-03-24T21:36:52Z

True, will do that

arichiardi · 2017-03-27T01:02:01Z

Done! 🎆 🎉 🎈

scottchiefbaker · 2017-03-29T21:46:11Z

This was with a brand new vanilla user on Fedora 25:

test@basement(~)
:sh install.sh
Check if '/home/test/bin' is writable: KO
Check if '/home/test/bin' is in PATH: OK
Check if '/usr/local/bin' is writable: KO
Check if '/usr/local/bin' is in PATH: OK
The default install directories could not be used.

In the first case, it looks like ~/bin is in the path by default, but not created? In that case I think it's a good idea to create the directory and put the files there. This was the behavior of the previous, and I think it's fine if ~/bin is in the path.

Lastly I'd like the install script to offer to set the recommended colors. diff-so-fancy --colors will output the commands for the recommended colors.

scottchiefbaker · 2017-03-29T21:47:04Z

Moving forward, I think this will be the recommended way to install d-s-f. Rather than cloning the whole repo, or using sytem repos. This will always ensure that the user gets the latest version via the master branch.

arichiardi · 2017-03-29T22:54:53Z

Yes I removed the creation part, assuming that if it is in the path, then it is writable/exists. But of course this assumption was wrong as it was shown above. I can bring it back to life.

scottchiefbaker · 2017-04-03T15:26:14Z

Yes please bring that logic back to life, if the directory is in the path and writable.

arichiardi · 2017-04-03T18:04:22Z

No well, if the directory is writable it means that it exists already...so only if it is in the path. Will do that.

scottchiefbaker · 2017-04-03T21:24:25Z

Question... Do we want to install to ~/bin/ by default... or require the use to specify?

bash install.sh /path/to/install/to/

Or both... if you call install.sh without a path it tries ~/bin and errors if it can't?

arichiardi · 2017-04-03T21:26:28Z

Yeah that is a UX question, I am fine with the current defaults, but I am going to wait for more answers to this one

scottchiefbaker · 2017-04-03T21:34:20Z

@paulirish @stevemao @OJFord

Feedback on this?

paulirish · 2017-04-03T21:43:26Z

install.sh

+dsf_do_install() {
+
+ if ! dsf_has dsf_download; then
+ echo >&2 "You need curl or wget to install nvm";


s/nvm/diff-so-fancy

paulirish · 2017-04-03T21:46:38Z

I really love the ~/bin/ convention and suggest keeping that. If someone doesn't want an easy install script, then they should be fine with our Manual Install directions.

It seems reasonable for this script to doublecheck that ~/bin is in your PATH though. Adding it to the PATH should definitely stay the user's job. Hate when scripts toss stuff into my .bash_profile. :)

arichiardi · 2017-04-03T21:50:32Z

@paulirish What about the creation of the folders if already in PATH? I usually don't like when scripts are messing with my stuff as well, but I assume that if you have something in PATH you actually want/wanted to have the folder in place and therefore I create it (maybe again if it had been removed).

On the other end if a user delete the folder but forgot to update PATH, creating it will maybe make her upset 😄

paulirish · 2017-04-03T21:53:16Z

oh i see you already do the PATH check. :) all good.

i think the way you currently handle it is best (is it writeable? is it in path? great, lets go)

paulirish

looks good to me.

can you also add something to the readme in this PR?
gives us an opportunity to bikeshed how it shows up there amongst the other options. :)

scottchiefbaker · 2017-04-03T21:58:04Z

@paulirish I'd like to completely revamp the README once we release 1.0.

It's a jumbled mess, with a ton of options.

paulirish · 2017-04-03T23:17:17Z

@scottchiefbaker oh word. yeah that sounds excellent! though i'd advocate for the revamp before shipping 1.0, assuming the release brings us some new attention. :)

arichiardi · 2017-04-03T23:39:11Z

Ups pushed the wrong version wait a sec

EDIT: ok worked on the other laptop on this, give me some time to get back to work and re-push the right version

arichiardi · 2017-04-04T17:24:29Z

Ok pushed the right version, still uncertain about reintroducing mkdir or not. Seeking advice.

arichiardi · 2017-07-30T09:44:55Z

What is the status on this? Is there anything I need to add/remove so that we can merge it? Good job with the release!

scottchiefbaker · 2017-08-02T16:44:36Z

I'm not sure where to go with this, I'm open to suggestions. With the release of 1.1.0 the installation is much simpler (one script). I think there is still a place for an install script that downloads the latest version and puts it in your $PATH appropriately. What do you think?

OJFord · 2017-08-02T16:59:42Z

an install script that downloads the latest version and puts it in your $PATH appropriately. What do you think?

Sounds like a package manager?

arichiardi · 2017-08-02T17:06:28Z

Oh well if there is one script already for that, and easy to call, then there is no need for an additional script.

scottchiefbaker · 2017-08-21T18:39:28Z

Now that things are really simple, I don't think we need a full blown install script. Maybe we just need to expand the install docs to cover some other use cases? I'm going to close this for now. If there are strong objections let me know and I'll re-open it.

arichiardi force-pushed the install-script branch from 94eaa4f to f40e674 Compare February 18, 2017 00:22

stevemao mentioned this pull request Mar 4, 2017

Diff-so-fancy needs usage output #223

Closed

scottchiefbaker added the enhancement label Mar 24, 2017

scottchiefbaker added this to the 1.0.0 milestone Mar 24, 2017

scottchiefbaker assigned arichiardi Mar 24, 2017

arichiardi force-pushed the install-script branch 2 times, most recently from 1659ae6 to e7f08d5 Compare March 27, 2017 01:00

stevemao referenced this pull request Mar 29, 2017

Bump version to 1.0.0

706e4f2

paulirish reviewed Apr 3, 2017

View reviewed changes

arichiardi force-pushed the install-script branch from e7f08d5 to ecf039a Compare April 3, 2017 23:36

Add fancy install script, closes so-fancy#217

2f6f7f7

arichiardi force-pushed the install-script branch from ecf039a to 2f6f7f7 Compare April 4, 2017 17:23

stevemao mentioned this pull request Jun 9, 2017

Create an official version 1.0.0 release #237

Closed

This was referenced Jun 26, 2017

Error locating diff-highlight #238

Closed

Integrate diff-highlight inline into d-s-f #226

Closed

stevemao removed this from the 1.0.0 milestone Jul 18, 2017

scottchiefbaker closed this Aug 21, 2017

Add fancy install script, closes #217 #218

Add fancy install script, closes #217 #218

Conversation

arichiardi commented Feb 18, 2017

arichiardi commented Feb 18, 2017

arichiardi commented Feb 18, 2017

scottchiefbaker commented Feb 28, 2017

arichiardi commented Mar 1, 2017 • edited Loading

OJFord commented Mar 1, 2017

arichiardi commented Mar 1, 2017

OJFord commented Mar 1, 2017

arichiardi commented Mar 1, 2017

scottchiefbaker commented Mar 1, 2017

scottchiefbaker commented Mar 1, 2017

arichiardi commented Mar 1, 2017 via email

OJFord commented Mar 1, 2017

scottchiefbaker commented Mar 1, 2017

scottchiefbaker commented Mar 6, 2017

arichiardi commented Mar 6, 2017

scottchiefbaker commented Mar 24, 2017

arichiardi commented Mar 24, 2017 via email

scottchiefbaker commented Mar 24, 2017

scottchiefbaker commented Mar 24, 2017 • edited Loading

scottchiefbaker commented Mar 24, 2017

arichiardi commented Mar 24, 2017

arichiardi commented Mar 27, 2017

scottchiefbaker commented Mar 29, 2017

scottchiefbaker commented Mar 29, 2017

arichiardi commented Mar 29, 2017

scottchiefbaker commented Apr 3, 2017

arichiardi commented Apr 3, 2017

scottchiefbaker commented Apr 3, 2017 • edited Loading

arichiardi commented Apr 3, 2017

scottchiefbaker commented Apr 3, 2017

paulirish Apr 3, 2017

Choose a reason for hiding this comment

paulirish commented Apr 3, 2017

arichiardi commented Apr 3, 2017

paulirish commented Apr 3, 2017

paulirish left a comment

Choose a reason for hiding this comment

scottchiefbaker commented Apr 3, 2017

paulirish commented Apr 3, 2017

arichiardi commented Apr 3, 2017 • edited Loading

arichiardi commented Apr 4, 2017

arichiardi commented Jul 30, 2017

scottchiefbaker commented Aug 2, 2017

OJFord commented Aug 2, 2017

arichiardi commented Aug 2, 2017

scottchiefbaker commented Aug 21, 2017

arichiardi commented Mar 1, 2017 •

edited

Loading

scottchiefbaker commented Mar 24, 2017 •

edited

Loading

scottchiefbaker commented Apr 3, 2017 •

edited

Loading

arichiardi commented Apr 3, 2017 •

edited

Loading