[Qgis-developer] Addressing a discrepancy between substr and strpos

[Mathieu Pellerin]

Sorry, the expression should have been
substr('1234;1234',*1*,strpos('1234;1234',';')),
which in QGIS returns '123'.

On Thu, Nov 19, 2015 at 1:33 PM, Mathieu Pellerin <nirvn.asia at gmail.com>
wrote:

> QGIS' expression engine has two string functions – substr() and strpos() –
> using two distinct character count standards. The substr() character count
> begins with 1, as per the equivalent postgresql string function. The
> strpos() character count however begins with 0, as per Qt / python's
> equivalent string function.
>
> This is a pretty ugly state of affairs as one expects those two functions
> to rely on the same character count method.
>
> For .e.g, here's a practical example: the expression
> substr('1234;1234',0,strpos('1234;1234',';')) would – in QGIS – return
> '123' because of the character count discrepancy between the two functions.
> That's very, very ugly and counter-intuitive.
>
> IMHO, substr()'s adherence to postgresql is the right way to go in the
> context of QGIS expressions, which means addressing this issue would mean
> fixing strpos() so its count starts at 1.
>
> Now, the question is, should – and if so how – we address this yuckiness.
> Should we leave things as it is, and add a note in the documentation about
> the discrepancy? Or should we fix strpos() acknowledging the longer we wait
> the harder it'll be, and make sure the changelog has a big bright note for
> users to know a fix has been applied?
>
> Awaiting opinions.
>
> Mathieu
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20151119/9db5808e/attachment.html>