RE: List of files modified between two tags

[Arthur Barrett]

Chaitanya,

> Hello Arthur...hope you're doing well. 

As well as can be expected after being on a flight for 23 hours...

> > I've written many an article here about how to go about 
> selecting a SCM
> > tool.
> >
> > An organisation installs SCM to achieve a buesiness 
> deliverable.  The
> > suitability of any tool should be measured against those 
> deliverables.
> > Anyone who tells you 'use tool xyz' or 'tool xyz is better than tool
> > abc' is just a zealot and not interested in helping you or your
> > organisation.  The right tool for the job can only be 
> determined until
> > the business drivers are determined.  
> >
> 
> Hello Arthur...hope you're doing well. The portion of your answer that
> I retained above is probably the most common doubt any newcomer to SCM
> will have. Could you please point me to the articles that you'd
> written previously on how to select an SCM tool?

Go to groups.google.com - and search gnu.cvs.help and also 
comp.software.config-mgmt

A couple of links to get you started:
http://groups.google.com/group/gnu.cvs.help/browse_thread/thread/185123331f17426c/7d8027d79becf3e6?lnk=gst&q=financial+times#7d8027d79becf3e6

http://groups.google.com/group/gnu.cvs.help/browse_thread/thread/f3e47c97e3e46e93/b57d93e1615458b2?lnk=gst&q=financial+times#b57d93e1615458b2

http://groups.google.com/group/gnu.cvs.help/browse_thread/thread/42e66908f42ce0bc/569418c98650d4b4?lnk=gst&q=financial+times#569418c98650d4b4


> 
> My question in detail:
> 
> How do you really classify a business deliverable? 

This is the point that a lot of 'developers' get stuck on.  Version control is 
not an end in itself - in fact SCM (software configuration management) is there 
to solve particular business problems like improve productivity, reduce costs, 
provide compliance with legislation etc.   It is possible to deliver a software 
project without version control (it happens all the time - even on large 
projects), so therefore version control in and of itself is not of interest to 
the business - but the productivity gains that SCM can bring may be of interest.

The bottom line with 'business deliverables' is that it must be measurable.  
Then you measure it now, and after your SCM system is in 
place/upgraded/whatever and you measure it again then you can tell if you have 
improved it.  Are bugs fixed faster?  Are fewer bugs found in delivered 
products?  Are customer satisfation levels higher?

> Is it along the
> lines of whether you deploy compiled code or scripted code? Or along
> the lines of programming languages used for the source? Or packaging
> procedures? 

No all of that is process - what process you follow to achieve the business 
deliverable.  But you are thinking well, because all these come before you 
select the tool, but before you answer any of these you must know the business 
driver.  Eg: the business wants to expand globally, including China, Korea, 
Arab states and Israel then the tools you use for software development (and 
SCM) will need to be proven to work well at storing and presenting information 
in those languages.  For instance: in our own testing we are yet to find any 
open source version control system that allows you to use Chinese login 
names/author names (except our own EVSCM).


> Suppose we deliver a big, heavy JPEG or something of a
> flowchart idea as our product.....or an RPM on Linux.....or a
> digitised movie.....or a PDF document...or GPS data in a CSV
> file...how does a deliverable like these influence your choice of an
> SCM tool? I've just used random examples but I hope you understood my
> perplexity here!

Yes this is a good point - but not the one I was making.  Certainly if your 
project has several 500Mb files then the final tools you use (including version 
control tools) must be capable of versioning these files.  I still think that 
the busines goal comes earlier than this - if the business needs SOX compliance 
and the VC tool that handles 500MB files well has no failsafe audit then it 
still wont deliver.

The thing to watch out for with these quasi-technical requirements is that 
different tools may resolve the problems in very different ways.  Eg: if you 
approached our sales team with a requirement to store 500MB files we'd want to 
know what was in them, and would probably recommend some way of storing the 
'source' and using a 'build' system to accurately rebuild the 500Mb file from 
easier to manage (smaller) sources.  This is common on Java projects with JAR 
files.  Similarly CM Synergy (previously called Continuus) is used by many 
large organistions and projects - but has a limitation of only a couple of 
hundred files in a single directory.  This limit does not stop it being used on 
large projects.

Money invested in getting consultants with proven expertise in this area to 
help you through the CM Design process is well worth it.  The company I work 
for does this, and this highlights a 'problem' with getting expertise in this.  
We also sell SCM tools, so there is a tendency to predispose any outcome to 
show our product is a 'fit', however the process we follow to get there should 
show the customer clearly that whilst our software is a 'fit' it is not 
exclusively the only 'fit'.  Also implementing SCM on a small scale and 
purchasing licenses for whatever software is needed is generally part of the 
'discovery' process.

Regards,



Arthur