That's fine with me. At this time, I'm really only concerned that we
have a flexible-enough API which has a CSS-like syntax to it.
Regards,
Matt
On Fri, Apr 18, 2008 at 1:30 PM, Andrew Robinson
<[EMAIL PROTECTED]> wrote:
> Could we leverage maven version parsing algorithms? Maven seems to be
> able to parse a lot of crap :)
>
> -Andrew
>
>
>
> On Fri, Apr 18, 2008 at 1:09 PM, Matt Cooper <[EMAIL PROTECTED]> wrote:
> > I agree, full agent "version" matching is troublesome. With this approach:
> >
> >
> > @agent ie and (min-major-version: 6) and (min-minor-version: 1) and
> > (max-major-version: 6) {
> > /* styles for IE agent version 6.1 through 6.x (inclusive) */
> > }
> >
> > We could go up to as many decimal tokens as we wish, e.g.
> > *-major-version, *-minor-version, *-maintenance-version,
> > *-build-version for "major.minor.maintenance.build" tokens.
> >
> > ----
> >
> > The question I would have then is what should we do with non-numeric
> > values like you pointed out; do we really care about skinning for
> > alphabetical versions?
> >
> > My proposal would be to take the raw version String and split it into
> > up into tokens based on the dot character. We should decide how many
> > tokens we want to support and then we should ignore any further tokens
> > in the String.
> > For each token, we turn it into an integer. If the token contains
> > something other than decimal characters, keep only up to the first
> > non-decimal character. Here are some examples:
> > a.) "0b5" would become "0"
> > b.) "b5" would become "" which we'd treat as "0"
> >
> > Regards,
> > Matt
> >
> > On Fri, Apr 18, 2008 at 12:55 PM, Andrew Robinson
> >
> >
> > <[EMAIL PROTECTED]> wrote:
> > > Perhaps matching the full agent string is a bad idea. I'd hate to have
> > > to parse many variations of things like:
> > >
> > > Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9b5)
> > > Gecko/2008032620 Firefox/3.0b5
> > >
> > > -Andrew
> > >
> > > On Fri, Apr 18, 2008 at 12:46 PM, Andrew Robinson
> > >
> > >
> > > <[EMAIL PROTECTED]> wrote:
> > > > The problem already brought up is that minor version may not be
> > > > enough. In a 3 part version (ie 2.0.10) the minor of 0 isn't helpful
> > > > if you want to do something based on the 10 value
> > > >
> > > > -Andrew
> > > >
> > > >
> > > >
> > > > On Fri, Apr 18, 2008 at 12:40 PM, Matt Cooper <[EMAIL PROTECTED]>
> wrote:
> > > > > Hi all,
> > > > >
> > > > > The regex would be powerful though I'm afraid that it would not
> as
> > > > > obvious or easy to use for non-technical designers/skinners.
> > > > >
> > > > > I think something like this would be clearer:
> > > > >
> > > > > @agent ie and (min-major-version: 6) and (min-minor-version: 1)
> and
> > > > > (max-major-version: 6) {
> > > > > /* styles for IE agent version 6.1 through 6.x (inclusive) */
> > > > > }
> > > > >
> > > > > Regards,
> > > > > Matt
> > > > >
> > > > > On Fri, Apr 18, 2008 at 12:30 PM, Andrew Robinson
> > > > >
> > > > >
> > > > > <[EMAIL PROTECTED]> wrote:
> > > > > > Well that is difficult isn't it? Perhaps what I suggested a
> long time
> > > > > > back in my bug is best and allow regexp:
> > > > > >
> > > > > > @agent blah and (matches-version: /someRegExp/)
> > > > > >
> > > > > > example to match 6.x through 7.x:
> > > > > > @agent ie and (matches-version: /[67](\.\d)*/)
> > > > > >
> > > > > > yeah it is harder to write, but then we can write it once and
> it
> > > > > > handles pretty much all use cases.
> > > > > >
> > > > > > The other twist is to give the code entire user agent string:
> > > > > >
> > > > > > @agent matches(/MSIE\s+[67]/)
> > > > > >
> > > > > > This really gives the user all the control they need and it
> is pretty
> > > > > > easy to parse without having to code many syntax improvements
> over
> > > > > > time as new requirements come up.
> > > > > >
> > > > > > -Andrew
> > > > > >
> > > > > > On Fri, Apr 18, 2008 at 12:10 PM, Andy Schwartz
> > > > > >
> > > > > > <[EMAIL PROTECTED]> wrote:
> > > > > >
> > > > > >
> > > > > > > On Thu, Apr 17, 2008 at 9:21 PM, Andrew Robinson
> > > > > > >
> > > > > > > <[EMAIL PROTECTED]> wrote:
> > > > > > >
> > > > > > > > I'll be happy either way, but I think I now bend to the
> below
> > > > > > > > explanation of 5 == 5.0 from Jeanne's reasoning
> > > > > > >
> > > > > > > FWIW, I actually agree with Jeanne's first opinion. :-)
> > > > > > >
> > > > > > > That is, I think of "5" as "5.*". "5.0" as "5.0.*", etc.
> > > > > > >
> > > > > > > Regarding the use of floating points to represent
> versions... I was
> > > > > > > wondering whether we should avoid this since it would
> prevents us from
> > > > > > > supporting "major.minor.reallyminor" version strings. I
> don't know
> > > > > > > that we will ever need to go further than major.minor,
> though the
> > > > > > > Gecko versions use the third digit, so perhaps we should
> pick a
> > > > > > > solution that doesn't preclude us from supporting this?
> > > > > > >
> > > > > > > (BTW, sorry all about my little digression earlier on the
> thread...)
> > > > > > >
> > > > > > > Andy
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
>
