Correct TC place for "debian mime" symlinks

[Yleisajattelija]

I have encountered installation problems with xsane backends and flashplayer. On both cases I think problem is this 'usr/' versus 'usr/local' installation point problem.  Actually these are not TC related problems, there is some bad code with absolute path referencies.

On 'xsane case' is maybe possible fix xsane backend source code, but on 'flashplayer' case I think there is no way to remove this problem. Reason for this is that '/usr' reference is probably done in HTLM stream. It comes from web page browsed, and the web page coder is making 'linux compatible' path reference with means RedHat and Debian style directory structure.

There is of course many other linux distroes, but it is impossible to web designer match code for 150 different linux distroes.

So, I will try to use symlinks to 'mime debian' -filestructure. Is that wise or/and possible goal?

If it is reasonable, what is best file/place to put these symlinks? I think many TC users use symlinks to solve these problems.

Question: is there any spesification (not of course, just asking) or common TC -practise for this?

[tinypoodle]

http://www.pathname.com/fhs/pub/fhs-2.3.html

[Yleisajattelija]

Tinypoodle, you are wright. FSH and LSB and beautiful, nice and correct things.

Unfortunately that helps nothing.

Indastry standard (and real, harsh, brutal world) is *.rpm and *.dpg -compatible linux packages.

And with those packages are RedHat and Debian filestructures, like it or not (and this is not my fault, I don't mean to be unpolite)

[althalus]

Tinypoodle, you are wright. FSH and LSB and beautiful, nice and correct things.

Unfortunately that helps nothing.

Indastry standard (and real, harsh, brutal world) is *.rpm and *.dpg -compatible linux packages.

And with those packages are RedHat and Debian filestructures, like it or not (and this is not my fault, I don't mean to be unpolite)

On 'xsane case' is maybe possible fix xsane backend source code, but on 'flashplayer' case I think there is no way to remove this problem. Reason for this is that '/usr' reference is probably done in HTLM stream. It comes from web page browsed, and the web page coder is making 'linux compatible' path reference with means RedHat and Debian style directory structure.

The local filepaths are almost definitely NOT defined in the HTML. That would be an absolute disaster security disaster.

There is of course many other linux distroes, but it is impossible to web designer match code for 150 different linux distroes.

As a web designer, I do not need to know or care which linux distro my visitors are coming from, as my code has absolutely no access to anything outside of the web browser. These days, I don't even need to care too much about what web browser my visitors are using, either, as long as it isn't IE6 (or to a lesser extent, newer versions of IE). Systems coders have a different experience, but that is an area well governed and managed by longstanding standards and best practices.

Indastry standard (and real, harsh, brutal world) is *.rpm and *.dpg -compatible linux packages.

No, industry standards is the link tinypoodle provided. Any binary packages provided that have hardcoded paths assuming a specific distro's layout are either:
1. doing it wrong
2. not designed to be run on other distros.

And for the record, Debian and Red Hat both implement the FHS.

If you need these incorrectly configured binary distributions running in your environment, then you must find ways around the issues. Manually creating symlinks is one such method, and should work effectively.

[Yleisajattelija]

The local filepaths are almost definitely NOT defined in the HTML. That would be an absolute disaster security disaster.

As a web designer, I do not need to know or care which linux distro my visitors are coming from, as my code has absolutely no access to anything outside of the web browser. These days, I don't even need to care too much about what web browser my visitors are using, either, as long as it isn't IE6 (or to a lesser extent, newer versions of IE). Systems coders have a different experience, but that is an area well governed and managed by longstanding standards and best practices.

No, industry standards is the link tinypoodle provided. Any binary packages provided that have hardcoded paths assuming a specific distro's layout are either:
1. doing it wrong
2. not designed to be run on other distros.

And for the record, Debian and Red Hat both implement the FHS.

If you need these incorrectly configured binary distributions running in your environment, then you must find ways around the issues. Manually creating symlinks is one such method, and should work effectively.

Althalus, you are right, too. And obviously you make good code and you have proper, healthy security principles.

But, as you mentioned there is lot of loose HTLM code on internet. Main reason is, that old IE version did not use strict code checking. But if Finnish state primary TV channel decides to implement some bad HTLM code on their internet video, there is very little I can do for it.

Same is with Xsane, as I know there is no alternative if I want use scanner with linux.

Lately I have read SamsungUnifiedDriver installation script code, and there is different installation if-elseif selection for every distro. TC is not specified, and script try to install 'linux universal driver' without success. Directories just do not match.

FHS or not, currently there is no universal way to install different distroes.