7704 2015-04-30 23:39:22 +0000 SRC_URI_append with override 2015-05-01 14:58:11 +0000 1 1 7 Build System, Metadata & Runtime BitBake bitbake 1.8.1 x86 Multiple RESOLVED NOTABUG Undecided normal --- 1 juro.bystricky richard.purdie poky.bs.watcher poky.watcher --- --- Don't know oldest_to_newest 50697 0 juro.bystricky 2015-04-30 23:39:22 +0000 Assume recipe contains the following and MACHINE=qemux86-64: SRC_URI = "file://file1" SRC_URI_qemux86-64 += "file://file2" SRC_URI_qemux86-64 += "file://file3" run bitbake -e and observe: SRC_URI="file://file2 file://file3" Seemingly identical program: SRC_URI = "file://file1" SRC_URI_append_qemux86-64 = "file://file2" SRC_URI_append_qemux86-64 = "file://file3" SRC_URI="file://file1 file://file2file://file3" So apparently the two syntaxes are not equivalent, and I suspect there is a bug in the first case. 50698 1 juro.bystricky 2015-04-30 23:41:32 +0000 It should say <snip> case 2: run bitbake -e and observe: SRC_URI="file://file1 file://file2 file://file3" So apparently the two syntaxes are not equivalent, and I suspect there is a bug in the first case. 50704 2 richard.purdie 2015-05-01 13:46:49 +0000 If you read the bitbake manual you will see that these are not intended to behave identically. This is how it is defined to behave. 50716 3 juro.bystricky 2015-05-01 14:58:11 +0000 (In reply to comment #2) > If you read the bitbake manual you will see that these are not intended to > behave identically. This is how it is defined to behave. Fair enough, RTFM. I got bitten by this but why blame self when I can blame it on a bug. However, I still think there are some inconsistencies. As I recall, this were the steps I took: Step 1: SRC_URI = "file://file1" SRC_URI += "file://file2" bitbake -> builds OK Step 2: "Improve" recipe by adding an override: SRC_URI = "file://file1" SRC_URI_qemux86-64 += "file://file2" bitbake -> builds OK Step 3: Final test, same "improved" recipe, clean build (wipe TMP) bitbake -> error: file1 not found.