lein deploy does not seem to honor :checksum :ignore

[sundbp]

I'm having trouble deploying a new snapshot to clojars, no matter what :checksum setting I set it always fail on checksum.

grep'ing through the code I don't see checksum being used in deploy.clj - perhaps it's only used for getting dependencies, not pushing deploys?

[michaelklishin]

Checksums are for dependencies you pull down. What error message do you get? What exactly are you running?

[sundbp]

lein deploy clojars =>
Retrieving tulos/bberg-sdk/3.6.1.0-SNAPSHOT/maven-metadata.xml (1k)
from https://clojars.org/repo/
Checksum validation failed, expected 42981a0f46d414b2a6577fcef74a3cc2b6fbbda4 but is da39a3ee5e6b4b0d3255bfef95601890afd80709
Retrieving tulos/bberg-sdk/3.6.1.0-SNAPSHOT/maven-metadata.xml (1k)
from https://clojars.org/repo/
Could not transfer metadata tulos:bberg-sdk:3.6.1.0-SNAPSHOT/maven-metadata.xml from/to clojars (https://clojars.org/repo/): Checksum validation failed, expected 42981a0f46d414b2a6577fcef74a3cc2b6fbbda4 but is da39a3ee5e6b4b0d3255bfef95601890afd80709
Failed to retrieve remote metadata tulos:bberg-sdk:3.6.1.0-SNAPSHOT/maven-metadata.xml: Could not transfer metadata tulos:bberg-sdk:3.6.1.0-SNAPSHOT/maven-metadata.xml from/to clojars (https://clojars.org/repo/): Checksum validation failed, expected 42981a0f46d414b2a6577fcef74a3cc2b6fbbda4 but is da39a3ee5e6b4b0d3255bfef95601890afd80709
Finished!

[technomancy]

I can't reproduce this problem. Maybe this jar has been redeployed to fix the checksums?

[technomancy]

Going to close this; if there's a repro case we can reopen.

[sundbp]

Yeah. Once the server side was patched up for the package in question could
not reproduce.
On 10 May 2013 00:30, "Phil Hagelberg" [email protected] wrote:

Going to close this; if there's a repro case we can reopen.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/1139#issuecomment-17696264
.

[sunng87]

I think we do need a command-line option to ignore checksum failure on deployment. In China the lein deploy timeouts frequently. Once it fails, a dirty checksum is left on the server that stops further retries.

[technomancy]

This seems like a way to address the symptoms instead of the cause of the problem.

[sunng87]

I found the checksum option in repositories works well for this issue. And the checksum option at root level is for dependencies only. Sometimes a command-line option for disable checksum is handy, because the connection issue only happens at some time. Perhaps I can create a PR for this feature?

[okal]

@technomancy Thanks for all the great work you do on leiningen.

Are you suggesting that the high latency Internet connections is the problem that needs solving in the case @sunng87 illustrated?

[hypirion]

@okal: I am guessing Phil meant that ignoring the warning is not the right answer – we/Clojars should probably handle incomplete deployments better.

I think the best solution would be to ensure that deployments are atomic – either they fail, or they succeed, not end up in some state in between.

[okal]

@hypirion Ah, that makes sense. Thanks for clarifying. Atomic deployments would be a great enhancement.

[hypirion]

Atomic deployments to Clojars is clojars/clojars-web#226. I'm pretty sure that is independent of Leiningen.