[CharlesS]

Okay, I have a shell script build phase in a target in an Xcode project. It runs at the end of the build process, copying the app bundle into a distribution folder and getting it packaged up nice and neat for distribution. This build phase ran great in Jaguar, but after upgrading to the new-style targets in Xcode, the thing is getting run at the same time as the linking phase (due to dual processors), resulting in an app bundle with no binary in the distribution folder.

I'm wondering if there's any way to get the shell script to wait until compiling is done before it does its thing. I'm interested in the "Inputs" and "Outputs" tables in the shell script's Info window, but I can't figure out how to add files to them. According to the docs,

The Inputs and Outputs tables let you specify the input and output files used by the script. The Inputs table specifies the files that the script operates on; the Outputs table specifies the files that the script produces. To add an entry to either of these tables, click in the table and type the name of the input or output file. Xcode determines whether to run the script by analyzing the input and output files. If the input files were modified more recently than the output files, Xcode executes the shell script. If you provide no outputs, Xcode runs the script every time you build the target.

However, clicking on the tables and typing does nothing.

What am I doing wrong?

[mikebeam]

A response i got about this on the xcode-users list from Apple, which worked for me:

"The bad news is the Shell Script Build Phase inspector is fairly broken in Xcode 1.0. The good news is that it is fixed in the next release.

With the current version you might be able to enter project-relative paths in the given table by clicking in the table, hitting return and then clicking in the new row to add the item."

[Arkham_c]

Originally posted by mikebeam:
A response i got about this on the xcode-users list from Apple, which worked for me:

"The bad news is the Shell Script Build Phase inspector is fairly broken in Xcode 1.0. The good news is that it is fixed in the next release.

With the current version you might be able to enter project-relative paths in the given table by clicking in the table, hitting return and then clicking in the new row to add the item."

Another workaround might be to "sleep" your shell script until the compiled output is present on the drive. Something like this:

Code:
sleep=true
while [ sleep ]; do
    if [ ! -f ${somefile} ]; then
        sleep 10
    else
        sleep=false
    fi
done

Untested, but it might work for you.

[CharlesS]

^ ^ I tried that quite a while ago, but it didn't work. What happens is that the [ ! -f somefile ] evaluates to false as soon as the file exists, but doesn't necessarily wait until the file is finished writing, so you end up working on a half-written file, and needless to say, you end up with something that doesn't work.

I ended up doing this, which does work:

Code:
while ps -cx | grep gcc
do
    sleep 1
done

Now that I know how to get something into the Inputs table (thanks, mikebeam!), I may try removing this part of the script and seeing if I can get it to work with the inputs and outputs rather than this somewhat kludgey method.