please caching not working

[Jeff Olsen]

I get a segmentation violation when trying to use tools/cache:http_cache_server

Steps to reproduce:

1) plz run tools/cache:http_cache_server

2) create .plzconfig.local with:

[cache]

httpurl = http://localhost:8080

httpwriteable = true

3) ./bootstrap.sh

Installing Go dependencies..................

Compiling built-in rules...

Building Please...

Building [388/573, 2.0s]:

=> [ 0.3s] //third_party/go:go-bindata Fetching...${ERASE_...

=> [ 0.1s] //third_party/go:humanize Fetching...${ERASE_AF...

=> [ 0.3s] //third_party/go:grpc Fetching... (14.1...

=> [ 0.5s] //third_party/go/zip:zip Compiling...${ERASE_AFT...

=> [ 1.8s] //third_party/go:gcfg Fetching...

=> [ 0.2s] //third_party/go:queue Fetching...

=> [ 0.0s] //third_party/go:semver Checking cache...${ERAS....

=> [ 0.8s] //third_party/go:warnings Fetching...${ERASE_AFT...

=> [ 0.1s] //third_party/go:openpgp Fetching...${ERASE_AFT....

=> [ 0.2s] //third_party/go:net Fetching...

panic: runtime error: invalid memory address or nil pointer dereference

[signal SIGSEGV: segmentation violation code=0x1 addr=0x40 pc=0x7d3c3b]

goroutine 1376 [running]:

cache.(*httpCache).StoreExtra(0xc42000c0c0, 0xc420530000, 0xc42063b4e0, 0x14, 0x14, 0xc420551100, 0x1d)

/home/jeff/src/commuity/please/src/cache/http_cache.go:67 +0x46b

cache.(*httpCache).Store(0xc42000c0c0, 0xc420530000, 0xc42063b4e0, 0x14, 0x14, 0xc89780, 0x0, 0x0)

/home/jeff/src/commuity/please/src/cache/http_cache.go:38 +0x12a

cache.cacheMultiplexer.storeUntil.func1(0xc420530000, 0xc42063b4e0, 0x14, 0x14, 0xc89780, 0x0, 0x0, 0xc420a30610, 0x9fa660, 0xc42000c0c0)

/home/jeff/src/commuity/please/src/cache/cache.go:78 +0x84

created by cache.cacheMultiplexer.storeUntil

/home/jeff/src/commuity/please/src/cache/cache.go:77 +0x115

exit status 2

Is http cache currently supported?

-Jeff

[Peter Ebden]

It's dropping behind the RPC cache in terms of features and definitely (as you can see!) in terms of testing.

I'll have a look at it; it might well get removed one day though, I'm not sure it offers any benefit over the RPC one really.

-P

[Peter Ebden]

FYI this should be fixed on master now - looks like a misplaced .Close on error.

[Jeff Olsen]

That did the trick, thanks!

-Jeff