Reply in-line :-

On Tue, Jan 27, 2009 at 17:35, sameer shinde s9sameer@gmail.com wrote:

Hi All,

Hi Sameer,

Whenever I visit squid-cache.org site. I always have a doubt is, why there are multiple stable version available for production release? What i mean is, when there is 3.0 stable release is available for production, then still why 2.7 is available. I can understand that, older versions should be available to the users, but then it may come under older version head not in the current versions list.

While I haven't worked with squid, I would just put up kernel wherever you have put up squid. For it follows the same pattern as you say of squid (there may be many more free software projects which follow the same model of development)

There are quite a few things which may be in upstream minds.

a. From what I understand, its never a good idea to have all the eggs in one basket. What happens if one release for some reason or the other bombs or doesn't function or has issue after being released. Software can never be termed stable as there are too many variables (hardware, different software versions etc.)

b. All the distributions do not wait for a specific release to take a version. Most of them have their own release cycles and sync with whatever software release is suitable for their users and has gone through QA .

c. Software regression is one of number of issues which plague software. http://en.wikipedia.org/wiki/Software_regression

Other thing is, when there is a latest version available, what is the need of develpment in the older version. i.e when we've ver3.0 then why 2.8. All the patches, bug fixes, new additions can be done in the newer release then why ver2.8?

Is there any specific reason for that? If yes, then which version one should use? Because all the versions give you production release.

Another factor to take into account is most of these releases have a long life.

For e.g. Intrepid Ibex (Ubuntu) uses the 2.6.27.x series. Now this was released in October 2008. This release has would be supported till April 2010.

Similarly, Jaunty has 2.6.28.x series which would need to be supported for 1.5 years as well.

While the bug-fixes may be needed or not, we don't know. Many a time due to other changes in technology they are able to tackle issues in a different way. Many a times bug-fixes flow from the latest version to the older version and not vice-versa. (although this is a possibility)

I also have observed that, even after publishing the stable version on site, they withdraw the older stable version, do something & again put back. My question is, if you withdraw any software version for any reason, it does not become a stable version. If anyone is doing some bug fixing into it or even a minor change of code, it should appear in the latest stable release. This gives you a track of bugs & bugfixes. But if you takeback any stable release and put it again after some time. What does it mean? If it was stable why it was taken back? If not does it really justify the stability?

Most of the projects have some sort of automated testing as well as some amount of human testing. Many people find it to be one of the tedious, unglamorous jobs so doesn't get much attention (as it should be).

For e.g. At this moment , the squid3.0 stable 10, stable3, stable 2 are withdrawn.

Some peoples find my questions silly, but I've a doubt and its always better ask instead of knowing nothing. And I'm also sure, many of ours must have asked this question to themself, but never been answered.

Share similar perspective

Of course, people who may have a better, more accurate answers please do so.

Sameer Shinde.
M:- +91 98204 61580
Millions saw the apple fall, but Newton was the one who asked why.

2009/1/27 sameer shinde s9sameer@gmail.com:

Hi All,

Whenever I visit squid-cache.org site. I always have a doubt is, why there are multiple stable version available for production release? What i mean is, when there is 3.0 stable release is available for production, then still why 2.7 is available. I can understand that, older versions should be available to the users, but then it may come under older version head not in the current versions list.

I see nothing wrong with having two current versions. I may have two servers, one which was installed last week with a fresh copy of CentOS5 and other which was installed 7 years back with Debian Woody. Running bleeding edge applications on older installations is a headache. There can be other issues like older hardware incapable of running latest and greatest software.

Anurag