Use a hashing function to construct output artfact file paths.

Modules like cpp need a way to construct unique locations
for output artifacts, but the current approach of duplicating
the path to the input file in the build directory can lead
to excessively long file paths that hit the limit on Windows.
So let's hash the path to the input file instead.

Change-Id: I47dcd242d0ee90c377f2a6df8e270bbefbaf082f
Reviewed-by: Thomas Epting <thomas.epting@stryker.com>
Reviewed-by: Joerg Bornemann <joerg.bornemann@theqtcompany.com>

1 files changed, 10 insertions, 0 deletions

diff --git a/doc/reference/jsextensions/jsextensions-general.qdoc b/doc/reference/jsextensions/jsextensions-general.qdoc
index 9582772fc..2bbd2f7ac 100644
--- a/doc/reference/jsextensions/jsextensions-general.qdoc
+++ b/doc/reference/jsextensions/jsextensions-general.qdoc
@@ -49,6 +49,16 @@
     Tries to find a variable with the given name in the build environment and returns its value. If
     no such variable could be found, \c undefined is returned.
 
+    \section2 qbs.getHash
+    \code
+    qbs.getHash(key: string): string
+    \endcode
+    Calculates a 16-byte hash of the input and returns it.
+    Rules in modules should use this function to find unique locations for output
+    artifacts in the build directory without duplicating the whole directory structure of
+    the respective input file (to deal with the case of two files with the same name in different
+    subdirectories of the same product).
+
     \section2 qbs.rfc1034Identifier
     \code
     qbs.rfc1034Identifier(str: string): string