Why I hate CVS tags

[Luigi Ballabio-2]

Hi all,
        annoyed as I am, I'll try and be pragmatic. So, let's compare figures:

a) How often did you actually look at the $Id$ cvs tag inside a source
    file? And how much time did you save by not doing "cvs status" instead?

b) How often having cvs tags into a file caused it to be touched upon
    committing it, thus tricking the makefile or the IDE into recompiling
    it again next time? (the file, and all dependent files, and all the
    examples and part of the test-suite...)
    How much time have you lost this way?

In my case, a/b << 1. Big time. So I say, down with $Id$ tags and stuff!
(hear, hear!) Let's remove the suckers, recompile the whole thing for the
last unneeded time, drink and be merry! (hear, hear!)

Thoughts?

Bye,
        Luigi

[Ferdinando M. Ametrano-2]

At 05:34 PM 5/15/2003 +0200, Luigi Ballabio wrote:
>         annoyed as I am, I'll try and be pragmatic. So, let's compare
> figures:
>[...]
>In my case, a/b << 1. Big time.
that a/b ratio is relevant only to those actively contributing to the CVS,
that is you and me. a/b << 1 for me too.

That said I think the $Id$ tag is worth the extra compile time, but since I
try to be pragmatic too I won't even argue with you on the subject, knowing
I would lose ;-)
Let's say that I can live without CVS tags.

I was way more sensitive on the quantlib.hpp inclusion in the test suite,
since to include quantlib.hpp is what is suggested to end user, and that
should be tested in my opinion. But you didn't asked my opinion on that ;-)


------------
ciao -- Nando

[Luigi Ballabio-2]

At 10:38 AM 5/16/03 +0200, Ferdinando Ametrano wrote:
>That said I think the $Id$ tag is worth the extra compile time, but since
>I try to be pragmatic too I won't even argue with you on the subject,
>knowing I would lose ;-)
>Let's say that I can live without CVS tags.

But you *won't* be living without tags! It's not like you don't get them if
you don't put the thing into the file! (wow, I just wrote a triple
negative) Ever tried doing "cvs status" on a file? You're even using
WinCVS---it shows you the tags right there in the directory listing...

>I was way more sensitive on the quantlib.hpp inclusion in the test suite,
>since to include quantlib.hpp is what is suggested to end user, and that
>should be tested in my opinion.

Ok, write a test case for that :) I'm not sure I understand what it would
be supposed to test, though (that quantlib.hpp exists? That the files it
includes exist?)

>But you didn't asked my opinion on that ;-)

Ouch...

Later,
         Luigi

[Ferdinando M. Ametrano-2]

>>I was way more sensitive on the quantlib.hpp inclusion in the test suite,
>>since to include quantlib.hpp is what is suggested to end user, and that
>>should be tested in my opinion.
>
>Ok, write a test case for that :) I'm not sure I understand what it would
>be supposed to test, though (that quantlib.hpp exists? That the files it
>includes exist?)
It should be tested that quantlib.hpp includes all (necessary) QuantLib
headers, and that the inclusion order does work. Minor issues, I agree, if
compared to the compile time implication


------------
ciao -- Nando