@mflatt @greg I’d like to have s3-sync also handle credentials from the environment. Currently it uses ensure-have-keys (which is deprecated). I could (1) change ensure-have-keys to also check the environment, or (2) change s3-sync to use credentials-from-environment!, but it’s important to keep all the things that currently work working. I think I’d prefer (1) (I would send a PR to the aws pkg) but I wanted to check with you first.
One aspect of this is that all the credential-related functions raise errors when they fail to find things, which makes composing them somewhat harder.
An alternative would be to have raco s3-sync take arguments to say how to configure authentication. An ensure-have-keys that infers available environment variables and files sure seems convenient, though; then again, I can see why explicit configuration is better in many contexts.
I could do that as well, but I’d prefer to have it just work — s3-sync is nice like that.
Also, if it did the “right thing” automatically, which should go first?
I image that environment variables should take precedence, and the legacy file format is obviously last.
@samth My brain has a cache miss on S3 credentials; would need some time to reload, to say anything much. For starters, could you rename-in ensure-have-keys, and define your own version in s3-sync? Maybe it uses with-handlers to try creds from env, else creds from file. (Also there’s two flavors of files, my old ad hoc, and the one later defined by AWS CLI tools, IIRC. Maybe for your purpose, only one is enough.) And if it’s all working as you like, decide if you want to stop there, or still make a PR? And if latter I could find time to reload brain cache and review.
I just wrote the (1 line of) code to handle an exception and fall back to ensure-have-keys in s3-sync. The real question is whether adding more “DWIM” to aws is desired.
I think from my point of view, ensure-have-keys is deprecated, and I’m not sure if I want to enhance it to dwim — esp if someday someone files a bug that the dwim only handles the legacy config file, not also the AWS CLI tools one, etc. etc. :slightly_smiling_face:
That’s my knee-jerk first reaction.
Ok
racket 7.7 release notes where :0?
But 7.7 hasn’t been released
Sounds dumb but how do I install ./ as a package?
I am passing --name but I still get raco pkg install: invalid package source;
ending path element is not a name
given: ./
for package name: softposit-herbie
Ah—don’t give a path at all!
It should work though. I was very confused when I attempted it and it didn’t work
I’ll probably submit a PR to fix this when I’m free
Does anyone have tips for looking at diffs of Racket code? git diff / line-oriented diff tools in general don’t produce the most satisfying output
I did a little bit of googling, but didn’t find any good alternatives
Though I don’t think it’s that readable
Tell diff not to compare whitespace helps some in my experience.
@me1150 could you share some examples of commits you made whose diffs were hard to read? Maybe there’s some formatting and code style changes that could help.
Re: https://github.com/zyrolasting/polyglot/issues/44, it looks as if the README.md in the source of polyglot was not distributed on a user’s Windows system. Is this due to a missing copy-shared-files binding in info.rkt, or is there another way I’m supposed to name static dependencies?