![](https://secure.gravatar.com/avatar/cfad84812bc41579778686925f4599d5.jpg?s=120&d=mm&r=g)
Hello, there is something which still puzzles me in \pdfliteral regarding q..Q rules. Lets take \def\test{\vrule width12bp height12bp depth0pt } To make some graphic-safe stuff in old pdfTeX one could to say \test \pdfliteral{q}% \test \pdfliteral{Q 1 0 0 1 12 0 cm}% \test Starting from pdfTeX 1.30 one can accomplish that using `page' keyword, which results in better content stream \test \pdfliteral page{q}% \test \pdfliteral page{Q}% \test But once we use both (which is not quite useless!) we finish somewhere outside the page; \test \pdfliteral page{q}% some graphic operators \pdfliteral{q}% some transform \test \pdfliteral{Q 1 0 0 1 12 0 cm}% \pdfliteral page{Q}% \test The reason is that last the matrix (translating coordinates back to the lower-left corner) is placed after the _second_ Q. But I'd rather expect that transform after the first (inner) one. What is the reason for that? -- Pawe/l Jackowski P.Jackowski@gust.org.pl