First documentation miss: Pertaining the mega manual's section 6.1.11. oe-init-build-env-memres, there are 2 references to a specific port "12345" when in fact it uses a randomly selected port if not specified. Second documentation miss: There is also this example of usage in that same section: $ source oe-init-build-env-memres ~/mybuilds Which fails for version 1.8 because this script currently consumes arguments in a positional manner and the first argument would be the port number instead of the builddir as implied. However, this statement might become true for 1.9 if the memres bitbake ends up being default as proposed in bug 7555. Third documentation miss: the bitbake-cookerdaemon.log file is not defined anywhere but at the toaster section. While it is good to make a reference there, the first and most complete definition of such log file should be in the same place where the memres bitbake exists. I filed this bug for the mega-manual but I'd assume this log file should also be described at the bitbake manual and I did not find it there.
Benjamin, All I could address was the "miss one" below. You can see the fix for that at http://www.yoctoproject.org/docs/1.8/mega-manual/mega-manual.html#structure-memres-core-script. For "miss two" - I don't understand how the user will even be able to enter that "source oe-init-build-env-memres ~/mybuilds" command if they were to have to provide a positionally first argument that was a randomly generated port address. If the address is random, how do they know in advance? So I am leaving this as is until I get some more explanation about it. For "miss three" - I cannot find a single occurrence of the "bitbake-cookerdaemon.log" string in either the most current mega-manual or the bitbake user manual. So I don't know where it is even being mentioned anywhere. Can you tell me specifically where you are seeing this log file mentioned in the docs? Thanks, Scott
A clarification was provided by phone to Scott for points 2 and 3 of this bug. Scott has all he needs now to fix this bug.
Benjamin, See http://www.yoctoproject.org/docs/1.8/ref-manual/ref-manual.html#structure-memres-core-script for the latest changes. I didn't really know what to say about that bitbake-cookerdaemon.log file. It is pretty brief and appears later in the section. If you have more details about the role that log plays then let me know and I can get more content in there around that file. Let me know on the other stuff how things look. Thanks, Scott
That looks like it addresses all the concerns. If you ask me, I would change this: The oe-init-build-env-memres script uses the bitbake-cookerdaemon.log file. This log is used exclusively for memory-resident build environments. For something like this: The _oe-init-build-env-memres_ script starts a memory resident bitbake server, this bitbake instance uses the _bitbake-cookerdaemon.log_ file which is located in the _build_dir_ Just to add a little info regarding why that file appears and where
Benjamin, I made that change verbatim. Thanks. I think we are good now. I am setting the bug to RESOLVED. See http://www.yoctoproject.org/docs/1.8/ref-manual/ref-manual.html#structure-memres-core-script Thanks, Scott
Setting documentation flag to "done"
The documentation for 1.9 was changed according to this bug, but for 1.8 the problem remains. http://www.yoctoproject.org/docs/1.8/ref-manual/ref-manual.html#structure-memres-core-script Put this bug on reopen.
Hi, The problem here was that the set of changes to the memres stuff spanned the point in time when fido was locked and tagged. So some of the changes did not actually make it into the 1.8 ref-manual. What I did here is was check out the fido branch, duplicated the complete set of changes to the ref-manual, rebuilt the manuals, and republished to the 1.8 area. This brings the 1.8 version of the ref-manual and the mega-manual, which includes the ref-manual, up to par. You can see the changes at http://www.yoctoproject.org/docs/1.8/ref-manual/ref-manual.html#structure-memres-core-script I am setting the bug back to RESOLVED/FIXED. Thanks, Scott
I fixed this and republished the 1.8 manual. I forgot to change the status of the bug to RESOLVED/FIXED. Status changed now.
Verified
Set the days to remove it from the report. The bug is VERIFIED/FIXED.