Don't Use UCM
This may be considered as
FAQ #0 of
(base) ClearCase FAQ.
It is
acknowledgedly controversial: consider it the pending of an implicit equivalent preliminary to the
UCM FAQ.
Here is a list of reasons to avoid UCM:
- UCM implements an obsolete concept.
- UCM is utterly a control more than a management tool. Using it, you become prisoner of decisions made upfront (restrictions on relocation, elements part of an original component, defined activities...).
- UCM itself is not meant to be manageable: it is a thick and opaque layer, with cryptic labels and complex, non-intuitive semantics. The user is clearly put in a subordinate, consumer's role. The need for and role of administrators is significant.
- UCM focuses on source elements exclusively—the concepts of software configuration, and of baseline, map to structured sets of sources. Derived objects are considered only second class citizens. There is no provision for using, in config specs, labels applied using configuration records, i.e. as the result of some auditing.
- UCM was obviously designed to support primarily GUI users, avoiding to put them in situation of having to make decisions, by enforcing already made ones. Unfortunately, it implemented an adverse scheme of publishing by merging back to integration branches, thus having to solve conflicts of resource acquisition, especially in presence of mastership.
- Question: But, UCM is based on activities, which are a higher-level concept (than branches)? Correct, but insufficient. It is not sure that activities are optimal in every case, and the UCM support for them brings excessive constraints, far outweighing their possible interest.
Practical details such as the lack of scalability, the poor performance, the complexity of the management and the lack of reversibility of many changes, are only consequences of these conceptual shortcomings.
I consider that the inception of UCM in the history of
ClearCase was a catastrophe, which stopped the development at an early stage, and nearly broke the conceptual integrity of this wonderful tool.
Of course, this leaves us with the question: if it is so bad,
why is it used at all?
--
MarcGirod - 04 Nov 2007