-
Notifications
You must be signed in to change notification settings - Fork 14
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
How to clean the cache during a build #1243
Comments
A tsc cache clean looks like it fixed the problem.
EDIT 8.9.2022 The steps to fix this are:
If successful, the output should be empty. Then open https://phet.colorado.edu/translate/trigger-build/number-play/{{locale}}/{{translatorId}} in a browser and logon. The locale and translator ID can be found by looking at the latest build-server logs, e.g.
|
@samreid is going to look at options for how we clean the cache during a build. |
In #1241 we observed that running a monolithic
I would add some timing notes from my mac M1: CCK AC: Here's the monolithic tsconfig I used for this test:
I'd rather not suffer another caching problem, so if we think |
|
Since there's nothing wrong with vegas or FiniteStatusBar, I'm going to retitle this issue to reflect what the actual problem is. |
We will probably investigate #1245 first. |
On hold until #1245 has progress. |
The original problem popped up on the build server again today. I reran the clean step manually. |
This happened again with number play again today, and #1245 is fixed. Bringing back up to dev meeting again. |
@zepumph: One fix is to blow away chipper/dist (either manually or with tsc --clean) every build on the build server. This would add a minute or two per build. The way we build and clean may change and so if our process changes we may need to update that here too, creating a maintenance burden. Just deleting chipper/dist would not have this problem, using @jonathanolson is concerned about the increase in time for batch maintenance releases. @mattpen should we do a maintenance release for number-play since that is the sim that seems to trigger this. @zepumph: Should we disable type checking for builds on the build-server? Lets continue this in a sub-group with @samreid @jonathanolson @zepumph. |
This happened again today. Updated manual steps to fix are in #1243 (comment). This takes me about an hour to fix every time it happens, it would be nice if we could maintenance release a fix. |
This happened again last night. I followed the steps in #1243 (comment), but there were additional errors (see below). I'm not quite sure how to fix these, so I wasn't able to get the build working. I restarted the build-server but number play is likely to keep failing. Can we please prioritize handling this in a sub group, @samreid @jonathanolson @zepumph?
|
Do we have a better option than just clearing chipper/dist? It seems like we might need to do that, and sim builds will take a bit longer. |
This happened again for the
I'm not sure, if that's what is needed I can add that to the build server. From what I understand that will make maintenance releases significantly longer. I recall @samreid saying something about not using the tsc cache in master? Maybe arithmetic just needs a maintenance release with an updated chipper sha? |
I reached out on slack to schedule a collaboration meeting. |
@mattpen @chrisklus and I were unable to reproduce this problem on the build server, even trying several sim/sim build combinations. So we made a maintenance release for number play 1.4 which avoids type checking in that chipper sha. Local testing seems OK. Closing until the problem recurs. |
Currently, number play is not able to build due to an error in vegas.
Steps to reproduce:
This only occurs on phet-server2 (running RHEL 8). Node version is 16.14.2. This does not occur on macOS 12 with Node 16.14.2.
grunt checkout-shas --repo=number-play
grunt
.Result:
The text was updated successfully, but these errors were encountered: