2010-10-14 18:17:42 How to handle warnings
T. Jump (UNITED STATES)
Message: 94521
I've been trying to understand issues that occur during a toolchain build (so I can include such issues in a module I'm writing for students). My process has been to work through errors that occur due to missing packages, then remove the toolchain and build it again to uncover what happens.
One thing I notice is different sets of warnings within each iteration of making a new build. From the document here ( docs.blackfin.uclinux.org/doku.php?id=optimization) it states, "A typical warning involves a variable being used before its value has been set."
So my question: is there a way to verify that the issue throwing the warning has cleared by the end of the make build?
And, is running make build again an option or could this cause problems within the build directories or image?
Thanks for any information you can share.
QuoteReplyEditDelete
2010-10-14 18:32:51 Re: How to handle warnings
Mike Frysinger (UNITED STATES)
Message: 94522
why are you building the toolchain ? there are precompiled ones already.
the set of warnings between toolchain builds should pretty much always be the same.
i really dont understand your question wrt verifying.
building the toolchain does not involve running `make` anywhere. are you sure you're actually building a toolchain ? or are you really building the distribution ?
post some actual commands you're running and the output that you dont understand.
QuoteReplyEditDelete
2010-10-14 22:50:58 Re: How to handle warnings
T. Jump (UNITED STATES)
Message: 94526
Thanks for the clarifying comments and questions. I'm building the distribution. I checkout an svn "sandbox"?? includes the distribution and some other code files I'm guessing), run make init, then make config, then make build.
The first few times I tried make build I had several errors that were resolved as I added new packages to my Ubuntu based system. Many times at first I never got an Images directory or any image files.
As I need to be able to reproduce the build successfully and predictably I'm capturing a list of all the needed group of packages and rerunning the checkout, make init, make config, make build until I can script the process (and get a deeper understanding of what is happening at each step).
Where I am in this process is I no longer get erros (I'm gaining confidence I have all the correct packages identified) but I do get warnings. It makes sense from the reference in my first post above that the warnings could be a result of values not yet processed, but what I'm looking for now is how to check to see if the values have been processed successfully ergo the warnings have been addressed/cleared. It would be great to have a process to run to show students the warnings are cleared. I noticed with my first few build tests (as I was clearing errors) that with each re-run of make build the number of warnings also decreased.
So, I'm looking to understand and be able to show the warnings being addressed, and if re-running make build is an acceptable process to do this or if there is another process I should use.
I've attached the warning output I now get with a single make build run for reference.
Hope this makes sense what I'm asking. Thanks.
make build 1 error report
QuoteReplyEditDelete
2010-10-14 22:58:40 Re: How to handle warnings
Mike Frysinger (UNITED STATES)
Message: 94527
why are you doing this trail & error for necessary packages ? why not simply use the documentation that already exists ?
docs.blackfin.uclinux.org/doku.php?id=installing_linux
as for your log capturing, using script like that prevents info from actually being useful. if you're going to use `script`, then simply use `make`. or forget this ancient `script` mumbo jumbo and redirect the whole output to a file:
make >& log
your log file does have an error, but because you've split things, there's no real way of knowing what package is causing it.
further, i'm not sure why you're doing `make build`. you should just be doing `make`.
QuoteReplyEditDelete
2010-10-14 23:28:48 Re: How to handle warnings
T. Jump (UNITED STATES)
Message: 94528
Preface: I picked up linux for the first time 18 days ago.
Trial and error for necessary packages was pre-being able to read the output and understand what I was seeing. Being completely new most of what I was reading in the available docs was "greek". Knowing which docs to read? What happens far too often is those in the know forget what it is like to know nothing and try to understand.
Also, much of the script scrolled off the top of the terminal window and I had no idea how to access it... until I learned how to redirect output with >. I could provide the full output (actually, I will attach it), but I was not asking for help in determining what was happening, but how to address if what was throwing warnings was cleared. If this can be determined by the full output then this helps answer my question which is how to I determine if the warnings have been satisfied. Let me know if this is the case and I will give it a go. FYI, I did not redirect the warnings/errors with > as they displayed in the terminal window (again, definitely more elegant ways to do this, but I still have much to learn. I don't feel compelled to appologize for not knowing; only not trying and for this I feel I have nothing to appologize for.
I did see the error... pretty sure I've solved this but I did not want to re-run make build (again, from the Getting Started document that came with the svn checked out file) until I got a handle on my issue of how to clarify warnings being satisfied.
Hope this answers your questions. Again, any help would be appreciated.
make build 1 script
QuoteReplyEditDelete
2010-10-15 02:45:10 Re: How to handle warnings
Mike Frysinger (UNITED STATES)
Message: 94541
well, that log doesnt show any errors ... so i guess you're all set ?
QuoteReplyEditDelete
2010-10-15 08:53:12 Re: How to handle warnings
T. Jump (UNITED STATES)
Message: 94562
I guess I'm not communicating very well. As I said before, I now understand how to find errors in the output (this is why I split the output with '> log' to generate the full log. But with only using '> log' the errors and warnings get parsed back to the terminal window). So the full output log of the distribution build would be the two documents I've attached above combined.
But when I deal with a bunch of students that start to get itchy about the warnings I want to have a complete answer as to why they occur and how to address them. To just say "ignore them" is not sufficient. I need to justify with evidence that they can be ignored. I'm just looking for a method to show students that the issues that threw the warnings have been addressed whether that be by finding evidence in the log or running a process to show they no longer appear.
OR... during the build does it keep warnings in a re-check pool somehow and the fact that no errors happen is the proof that the warnings were finally addressed? I'm just trying to understand how to know that the issues that throw warnings are fixed or addressed or why they can be ignored.
Hope this is clear as to the information I'm looking for.
QuoteReplyEditDelete
2010-10-16 15:42:25 Re: How to handle warnings
Mike Frysinger (UNITED STATES)
Message: 94596
we dont keep track of the warnings generated. and i imagine some of those warnings are important and should be fixed. but you're talking about hundreds of open source packages that we dont have anything to do with. we have no resources to look into every random package and fix their problems. we focus on the ones that actually affect people (and we know those ones when customers hit runtime problems and complain).
QuoteReplyEditDelete