I’m running into an unexpected problem with the package build server. I created libsass yesterday, which distributes pre-built versions of libsass for mac and linux and, instead of using git as the package source, I used an archive hosted on one of my servers. The package server reports the expected checksum as 8c9f5c0f0fdb6862dc6b3c5a86cd28452af293df, which is what my *.CHECKSUM file contains, but the build server reports:

pkg: mismatched checksum on package
  package source: <https://racket.defn.io/racket-libsass-3.6.1.tar.gz>
  expected: "23613da41a38e7b0ebb76a68c0a724612ee1b76f"
  got: "8c9f5c0f0fdb6862dc6b3c5a86cd28452af293df"

But if I grab the package from the server and run sha1sum on it, I get:

root@3e31274f96d8:/# sha1sum racket-libsass-3.6.1.tar.gz
8c9f5c0f0fdb6862dc6b3c5a86cd28452af293df  racket-libsass-3.6.1.tar.gz

root@3e31274f96d8:/# raco pkg install --name libsass <https://racket.defn.io/racket-libsass-3.6.1.tar.gz>
Downloading checksum for libsass
Downloading <https://racket.defn.io/racket-libsass-3.6.1.tar.gz>
raco setup: version: 7.4
raco setup: platform: x86_64-linux-natipkg [3m]
raco setup: target machine: racket
raco setup: installation name: 7.4
raco setup: variants: 3m
raco setup: main collects: /usr/share/racket/collects
raco setup: collects paths:
raco setup:   /root/.racket/7.4/collects
raco setup:   /usr/share/racket/collects
raco setup: main pkgs: /usr/share/racket/pkgs
raco setup: pkgs paths:
raco setup:   /usr/share/racket/pkgs
raco setup:   /root/.racket/7.4/pkgs
raco setup: links files:
raco setup:   /usr/share/racket/links.rktd
raco setup:   /root/.racket/7.4/links.rktd
raco setup: main docs: /usr/share/racket/doc
raco setup: --- updating info-domain tables ---                    [7:25:29]
raco setup: updating: /root/.racket/7.4/share/info-cache.rktd
raco setup: --- pre-installing collections ---                     [7:25:29]
raco setup: --- installing foreign libraries ---                   [7:25:29]
raco setup: --- installing shared files ---                        [7:25:29]
raco setup: --- compiling collections ---                          [7:25:29]
raco setup: --- parallel build using 4 jobs ---                    [7:25:29]
raco setup: 3 making: &lt;pkgs&gt;/libsass
raco setup: 3 making: &lt;pkgs&gt;/libsass/artifacts
raco setup: --- creating launchers ---                             [7:25:31]
raco setup: --- installing man pages ---                           [7:25:31]
raco setup: --- building documentation ---                         [7:25:31]
raco setup: --- installing collections ---                         [7:25:32]
raco setup: --- post-installing collections ---                    [7:25:32]

… and raco seems to happily install it in this docker container

@popa.bogdanp no idea why the checksum stuff is being weird, but just wanted to say that using GitHub Actions to build the libsass packages is a super neat idea

Thanks! I ran out of time yesterday, so I gave up on the idea temporarily, but I’ll come back to it. Ideally, GH Actions would build the artifacts and commit them to the repo (stored using git lfs), but I haven’t checked if raco pkg supports lfs. I assume it doesn’t so what I think I’ll end up doing is have GH Actions build the artifacts and then generate the final tar file and ship it to my server.

Does anyone have the hirez graphic file for the [:racket-flat:Racket] (logo and name) sticker on devswag? https://devswag.com/products/racket

Tia Stephen

[Getting stickers sent outside the US postal system adds and extra $16 to the price at devswag, so I’ve found local sticker printers but while I have the hires logo artwork (on wikipedia), I don’t have the logo&text artwork :white_frowning_face: - I used redbubble for the Australians, and am working out what do do for the European winners]

Still on this - sorry it’s delaying racket#2808 - is there a way to see which absolute path is found for the dll ? I have probably around 15 instances of this file on my C: with different versions.

So, on windows dll discovery follows %PATH% it seems. I should be able to find the bad/old dll now, I guess.

You are right. It must be a continuation of the example that begins: > After mastering the world of blogging software, you decide to put the ubiquitous http://Add-Two-Numbers.com\|Add-Two-Numbers.com out of business with http://Sum.com\|Sum.com:

@popa.bogdanp Just summing up. The build-server gets the correct file with the correct checksum, but for some reason, it expects a wrong checksum. That is, it ignores your checksum file?

The way I interpret that error is its own sha of the tar file was “23613da41a38e7b0ebb76a68c0a724612ee1b76f”, and my checksum file contained “8c9f5c0f0fdb6862dc6b3c5a86cd28452af293df”

The latter is the real checksum of the file so I’m not sure how the package server got “23613da41a38e7b0ebb76a68c0a724612ee1b76f”.

Maybe it was interrupted in the middle of downloading the file so it ended up with a truncated file

pkg: mismatched checksum on package
  package source: <https://racket.defn.io/racket-libsass-3.6.1.tar.gz>
  expected: "23613da41a38e7b0ebb76a68c0a724612ee1b76f"
  got: "8c9f5c0f0fdb6862dc6b3c5a86cd28452af293df"

Sure it is not the other way around?

Hmm. Depends on the perspective of the build server.

Looks like you’re right:

$ echo "foo" > dist/racket-libsass-3.6.1.tar.gz.CHECKSUM
$ raco pkg install --name libsass2 dist/racket-libsass-3.6.1.tar.gz
raco pkg install: mismatched checksum on package
  package source: dist/racket-libsass-3.6.1.tar.gz
  expected: "foo\n"
  got: "8c9f5c0f0fdb6862dc6b3c5a86cd28452af293df"

however:

$ curl <https://racket.defn.io/racket-libsass-3.6.1.tar.gz.CHECKSUM>
8c9f5c0f0fdb6862dc6b3c5a86cd28452af293df

so that’s even more bizarre

Is this were the checking happens?

https://github.com/racket/racket/blob/0257b5cf71fd3c53b75cfde5ef4d27052eef2a57/racket/collects/pkg/private/stage.rkt#L267

My current theories are that it either

• saved the checksum that I had uploaded when I first created the package on the package server and then tried to compare that saved checksum to an updated version of the archive that I later uploaded, without retrieving the new checksum file, or • it tried to build the package while I was uploading an updated version of the archive.

The latter doesn’t seem that likely since, afaict, package builds run late in the afternoon my time and the last time I uploaded the package was around noon. Either way, I’ve edited the package and I’m hoping that’ll trigger a new build in a few hours so we’ll see what happens then.

You could probably use GH Releases to host the tarball

So Racket find the ssleay.dll distributed with Intel iCLS, which is really old. :disappointed: Annoying. Should/Could we do a DLL version check before using it for ffi? It should be technically possible, but I have never played much with Windows dll stuff.

Good idea. I’ll have to give that a try

Well, this is a build-specific problem, since Racket always looks in its own “lib” directory first. Building Racket on Windows is difficult enough without trying to accommodate hostile installations… :slightly_smiling_face:

A way to make progress on the build is to manually put the right “ssleay32.dll” and “libeay32.dll” in the “racket/lib” directory. You can get them here, if needed: https://github.com/racket/libs/tree/master/racket-win32-x86_64-3/racket

Is this1 a reasonable use of expand-import? I’m trying to export specialized versions of certain functions from a couple of modules and I would prefer to not have to list them all. It seems to work fine, but is it going to come back to bite me in the butt later?

I think that’s reasonable, but …

could you instead have provide-easy-version be (provide (easy-version ...)) by using a provide pre-transformer and then just use except-out instead of defining that macro at all?

I’ll give that a try

Did that - now working. Thanks.

Does anyone know if we actually have the rights for the images used in 2htdp/planetcute?

As far as I can tell, the documentation just links to a (mostly broken) old page from the author that doesn’t list any sort of license.

And the Racket source code doesn’t seem to have a license.

(for those images)

@spdegabrielle Umm…. @ben might have some off hand?

@leif I believe they originate from here: http://www.lostgarden.com/2007/05/dancs-miraculously-flexible-game.html

@spdegabrielle You can also find svg images of it (and some past ones) at https://users.cs.northwestern.edu/~robby/logos/

thanks @robby ^.^

@soegaard2 Yup, they do.

(At least according to the link in the docs)

But there still isn’t any license attached on that page. (At least not any that I’ve found.)

@leif This page has a paragraph “The License”: http://www.lostgarden.com/search/label/free%20game%20graphics

And here it is: http://www.lostgarden.com/search/label/License

@leif the @robby logos page doesn’t include the image for the ‘racket’ sticker

@soegaard2 ah, okay, thanks.

(Oey, almost every link on that website seams broken, and just takes me to the home page.)

Yeah, very annoying. I attempted to download the zip-file with the graphics. A no go.

I see.

Also, the Basic Licensing and Cloning sections seem to contradict each other. :confused:

!

Does the statement on http://www.lostgarden.com/p/contact.html?m=1 help?

The problematic bits being the “All licensed items use the Creative Commons Attribution 3.0 License.” from the first section, and “You are expressly forbidden to use my graphics in a clone of one of my commercial games.” in the second.

@spdegabrielle Sadly nah. That seems to point to the same page as @soegaard2’s search pops up. :confused:

It does say: > I no longer work as a contract artist and haven’t for well over a decade. I do however provide several sets of ancient artwork intended for use by indies. These are provided as is. I will not be updating them, but you are welcome to mash them up as desired.

I mean, it probably isn’t a problem for racket….as ccy-by is (iirc) compatible with the lgpl.

Depending on how that ‘cloning’ clause plays out anyway.

I suppose he changed the original “license” from “don’t use it commercially” to “CCA 3.0” at some point.

I’m hoping to organise a gamejam at some point.

He should update his site.

@spdegabrielle Cool.

@soegaard2 Ideally yes. Although I get the impression he doesn’t care enough to do that.

Planet-cute was on the list of resources

@spdegabrielle I see.

@spdegabrielle On your racket sticker question, do you know what font that is?

It looks like source code pro to me.

https://fonts.google.com/specimen/Source+Code+Pro

If so, it should be pretty trivial to put the SVG lambda next to the word Racket written in that font.

I thought it might be a MB font?

Like, I could quickly do it if that helped.

It might be. Not sure which though.

I thought it looked good. (Good enough to adorn my old laptop :grinning:)

I don’t trust myself to do as nice a job- so if you could do a new version that would be great.

Sure!

Although I will ask @ben if he has a copy first. :slightly_smiling_face:

@soegaard2 Going back to planet cute, if only I could find a zip of those assets. Presuming it had a license in it, that would make it much clearer what the license is.

Let’s check http://archive.org\|archive.org

https://web.archive.org/web/20070715112638/http://lostgarden.com/PlanetCute%20PNG.zip

Yup, that has it. Sadly he didn’t put a license in that zip file. :disappointed:

There doesn’t appear to be anyone with the ability to change the topic in the IRC channel anymore. It’s stuck on Racket 7.3

Done. sorry, I accidentally got disconnected.

@spdegabrielle @leif here are the two SVG files I used for stickers (circle + rectangle)

Thanks @ben

Does that work for you @spdegabrielle?

nice! thanks @ben @leif