Bug 125682 – Post-process exported WebKit headers based on PLATFORM_NAME

• Component: New Bugs
• Status: RESOLVED FIXED

[Andy Estes]

Post-process exported WebKit headers based on SDKROOT

[Andy Estes]

Created attachment 219161 [details]
Post-process exported WebKit headers based on SDKROOT

[Andy Estes]

Here's a formatted version of the script:

postProcessInDirectory() {
    cd "$1"

    shopt -s nocasematch;
    if [[ ${SDKROOT} =~ iphoneos ]]; then
        local unifdefOptions="-DTARGET_OS_EMBEDDED=1 -DTARGET_OS_IPHONE=1 -DTARGET_IPHONE_SIMULATOR=0";
    elif [[ ${SDKROOT} =~ iphonesimulator ]]; then
        local unifdefOptions="-DTARGET_OS_EMBEDDED=0 -DTARGET_OS_IPHONE=1 -DTARGET_IPHONE_SIMULATOR=1";
    else
        local unifdefOptions="-DTARGET_OS_EMBEDDED=0 -DTARGET_OS_IPHONE=0 -DTARGET_IPHONE_SIMULATOR=0";
    fi

    for header in $(find . -name '*.h' -type f); do
        unifdef -B ${unifdefOptions} -o ${header}.unifdef ${header}
        case $? in
        0)
            rm ${header}.unifdef
            ;;
        1)
            mv ${header}{.unifdef,}
            ;;
        *)
            exit 1
            ;;
        esac
    done
}

postProcessInDirectory "${TARGET_BUILD_DIR}/${PUBLIC_HEADERS_FOLDER_PATH}"
postProcessInDirectory "${TARGET_BUILD_DIR}/${PRIVATE_HEADERS_FOLDER_PATH}"

(In reply to comment #2)
> Here's a formatted version of the script:
> 
> postProcessInDirectory() {
>     cd "$1"
> 
>     shopt -s nocasematch;
>     if [[ ${SDKROOT} =~ iphoneos ]]; then

SDKROOT is a path which may contain arbitrary components, any of which may match “iphoneos”. Could we use PLATFORM_NAME here?

[Andy Estes]

(In reply to comment #3)
> (In reply to comment #2)
> > Here's a formatted version of the script:
> > 
> > postProcessInDirectory() {
> >     cd "$1"
> > 
> >     shopt -s nocasematch;
> >     if [[ ${SDKROOT} =~ iphoneos ]]; then
> 
> SDKROOT is a path which may contain arbitrary components, any of which may match “iphoneos”. Could we use PLATFORM_NAME here?

Good point. Yes, I can use PLATFORM_NAME instead.

[Andy Estes]

Created attachment 219176 [details]
Post-process exported WebKit headers based on PLATFORM_NAME

[Andy Estes]

(In reply to comment #5)
> Created an attachment (id=219176) [details]
> Post-process exported WebKit headers based on PLATFORM_NAME

Here's the updated script with formatting:

postProcessInDirectory() {
    cd "$1"

    shopt -s nocasematch;
    if [[ ${PLATFORM_NAME} == iphoneos ]]; then
        local unifdefOptions="-DTARGET_OS_EMBEDDED=1 -DTARGET_OS_IPHONE=1 -DTARGET_IPHONE_SIMULATOR=0";
    elif [[ ${PLATFORM_NAME} == iphonesimulator ]]; then
        local unifdefOptions="-DTARGET_OS_EMBEDDED=0 -DTARGET_OS_IPHONE=1 -DTARGET_IPHONE_SIMULATOR=1";
    else
        local unifdefOptions="-DTARGET_OS_EMBEDDED=0 -DTARGET_OS_IPHONE=0 -DTARGET_IPHONE_SIMULATOR=0";
    fi

    for header in $(find . -name '*.h' -type f); do
        unifdef -B ${unifdefOptions} -o ${header}.unifdef ${header}
        case $? in
        0)
            rm ${header}.unifdef
            ;;
        1)
            mv ${header}{.unifdef,}
            ;;
        *)
            exit 1
            ;;
        esac
    done
}

postProcessInDirectory "${TARGET_BUILD_DIR}/${PUBLIC_HEADERS_FOLDER_PATH}"
postProcessInDirectory "${TARGET_BUILD_DIR}/${PRIVATE_HEADERS_FOLDER_PATH}"

[Andy Estes]

(In reply to comment #6)
> (In reply to comment #5)
> > Created an attachment (id=219176) [details] [details]
> > Post-process exported WebKit headers based on PLATFORM_NAME
> 
> Here's the updated script with formatting:
> 
>     shopt -s nocasematch;

Oops. This line can be removed now.

Comment on attachment 219176 [details]
Post-process exported WebKit headers based on PLATFORM_NAME

r=me
How long does this step take?

[Andy Estes]

(In reply to comment #8)
> (From update of attachment 219176 [details])
> r=me

Thanks!

> How long does this step take?

When no headers require post-processing, the step takes ~1.2s. With a clean build in a tree containing the iOS port's modifications to WebKit headers, the step takes ~1.4s. This is on a Late 2012 13-inch MacBook Pro with Retina display.

[Andy Estes]

Committed r160559: <http://trac.webkit.org/changeset/160559>