IMHO no more info than @Florjon gave is needed. Maybe some small details are left to understand why it might not work for us sometimes.
First of all, the 

(hex) or 

(dec) inside a <xsl:text/>
will always work, but you may not see it.
<br/>
will do fine. Otherwise you'll see a white space. Viewing the source from the browser will tell you what really happened. However, there are cases you expect this behaviour, especially if the consumer is not directly a browser. For instance, you want to create an HTML page and view its structure formatted nicely with empty lines and idents before serving it to the browser.disable-output-escaping
and where you don't. Take the following example where I had to create an xml from another and declare its DTD from a stylesheet. The first version does escape the characters (default for xsl:text)
<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
<xsl:output method="xml" indent="yes" encoding="utf-8"/>
<xsl:template match="/">
<xsl:text><!DOCTYPE Subscriptions SYSTEM "Subscriptions.dtd">


</xsl:text>
<xsl:copy>
<xsl:apply-templates select="*" mode="copy"/>
</xsl:copy>
</xsl:template>
<xsl:template match="@*|node()" mode="copy">
<xsl:copy>
<xsl:apply-templates select="@*|node()" mode="copy"/>
</xsl:copy>
</xsl:template>
</xsl:stylesheet>
and here is the result:
<?xml version="1.0" encoding="utf-8"?>
<!DOCTYPE Subscriptions SYSTEM "Subscriptions.dtd">
<Subscriptions>
<User id="1"/>
</Subscriptions>
Ok, it does what we expect, escaping is done so that the characters we used are displayed properly. The XML part formatting inside the root node is handled by ident="yes"
. But with a closer look we see that the newline character 

was not escaped and translated as is, performing a double linefeed! I don't have an explanation on this, will be good to know. Anyone?
The second version does not escape the characters so they're producing what they're meant for. The change made was:
<xsl:text disable-output-escaping="yes"><!DOCTYPE Subscriptions SYSTEM "Subscriptions.dtd">


</xsl:text>
and here is the result:
<?xml version="1.0" encoding="utf-8"?>
<!DOCTYPE Subscriptions SYSTEM "Subscriptions.dtd">
<Subscriptions>
<User id="1"/>
</Subscriptions>
and that will be ok. Both cr and lf are properly rendered.
nl
, not crlf
(nl=lf
). My first attempt was to use only cr:
and while the output xml was validated by DOM properly. I was viewing a corrupted xml:
<?xml version="1.0" encoding="utf-8"?>
<Subscriptions>riptions SYSTEM "Subscriptions.dtd">
<User id="1"/>
</Subscriptions>
DOM parser disregarded control characters but the rendered didn't. I spent quite some time bumping my head before I realised how silly I was not seeing this!
For the record, I do use a variable inside the body with both CRLF just to be 100% sure it will work everywhere.