Is Clearcase recursive labelling more efficient than element by element labelling?

Question

I use Clearcase on a project with ~3700 files. While making a release, there are some directories that have to be labeled recursively. The file count here is ~2400. The files in other directories are selectively labelled. I currently label the files using a script to iterate through a list of files and label the files one by one. This takes around 20 mins while connected to Ethernet.

Is it efficient to split the labeling process into two parts one for recursive labelling and one for selective labeling? Is recursive labeling faster by 1x,2x,..?

Answer 1

ClearCase operations file by file are always... slow!

You need to apply your label recursively if you can (that is if all the files of a given tree need labeling).
It is also recommended to do that operation on a dynamic view, in order to avoid any side effect related to an update status of a snapshot view (not updated, or update in progress)

The result will be faster (I do not have an exact estimation, but at least 2x seems right).

Warning, your directory from which you apply recursively the label must be in the right version (i.e. the version selected by the config spec).

---

Do not forget that the point of labeling is to identify a coherent set of file (i.e. a set of file which evolves and is labeled as a all). That means "mklabel -rec" is always better than putting a label on a single file.
A recursive label does not miss any file, a label put on files from a list can result in an incomplete set (for instance, if the list of files to label is obsolete or incomplete)

Answer 2

Why don't you use the Apply Label tool? That's what we do anyway.