Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Why not switch to Ubuntu-style version numbers? If they have to ship multiple version in a month, either include the day in the version number or call it Update X.

Using the date for the version number gets rid of any connotations associated with versioning (a .1 release being smaller than a whole number release, or that Firefox 7 is less advanced than Chrome 15 because its version number is half of Chrome's) while still allowing users to identify what version they're running.



A sequential version number gives a [probably misleading] indication of maturity.

A chronological version number gives an indication of currency.

Hiding the version number doesn't seem helpful.


How do you use date-based version numbers for planning purposes? Do you have another versioning system just to list what will be included in each release?


With a rigid release schedule, it's easy to figure out the release date for a given version. Luckily, Mozilla has already standardized on development cycles of 18 weeks (http://blog.mozilla.com/futurereleases/2011/07/19/every-six-...) staggered to result in releases every 6 weeks.


Distant Ubuntu releases (not yet codenamed) are generally referred to as [latest_codename] + x, like "Compositing will actually work well in Natty+5", meaning the feature is targeted for five releases after Natty.

Since Firefox has "Aurora" and "Beta" channels on a regular graduation schedule (Trunk->Aurora, Aurora->Beta, Beta->Final), the only question is "When is unmerged feature X going to be placed in trunk?", and the answer to that is usually "when we expect it to work well enough".

What will be interesting here is when there are MAJOR changes that take more than six weeks of testing to stabilize. Will the cycle just stall temporarily and go immediately back to six weeks on the next cycle?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: