Hairline strokes (linewidth 0) in imported PDFs are not visible

- reproduced on Windows 7, Inkscape 0.48.4

- not reproduced on Windows XP, Inkscape development rev 13031, result attached.
- which OS and Inkscape are you running? Depending somewhat on your OS, you may be able to try a more recent development version of Inkscape.

I'm using 0.48.4 r9939 on win7.

I just tested the 0.48+devel r13012 version and the problem is solved. Thanks.

I just hope inkscape releases a stable version soon. 0.48.4 is getting old. :-)

good to hear, in that case I'll mark this as Fix Committed, without knowing exactly when this was fixed.

wrt to a new release, I know a new release is in the works, but not sure of the precise schedule:

http://inkscape.13.x6.nabble.com/Bug-Hunt-for-0-91-td4969141.html

Based on tests with archived builds (reproduced with rev <= 11361, not reproduced with rev >= 11364), this was fixed in r11364:
<http://bazaar.launchpad.net/~inkscape.dev/inkscape/trunk/revision/11364>

See also:
<http://thread.gmane.org/gmane.comp.graphics.inkscape.user/14539>

Hello,

I found another file that reproduces the problem even with recent inkscape devel version r13012.

The graph on page 2 (and others). Also, the table on page 4 is imported with different line width while in adobe reader, there is no visual difference.

Backported to stable in r10015

    The problem encountered in the file Artigo VII CITENEL is somewhat similar to the problem originally fixed for file Graph LO.pdf, but it is more complex this time. In the original problem, fixed in rev 11364, the problem was just that the linewidth did not have a default value. Now it has a default value of 1. The new problem shown in file Artigo VII CITENEL, page 2, graph called Tempo, is that the linewidth is not initiallized, and also that it is embedded in a transformed group with a scale factor of 0.014 roughly, so the effective linewidth is 0.014 even after initiallization, which is not easily visible. Typical group element is shown below.
    This will be much more difficult to solve than the original bug report.

...................................................

<g id="g3572" transform="matrix(0.014108,0,0,-0.0136098,303.55994,770)">
    <g id="g3574" transform="translate(16539,1432)">
        <path d="M 0,0 -13336,0"
              style="fill:none;stroke:#b4b4b4;stroke-width:1;stroke-linecap:square;stroke-linejoin:miter;
              stroke-miterlimit:10;stroke-opacity:1;stroke-dasharray:none"
              id="path3576"
              inkscape:connector-curvature="0" />
    </g>
</g>

What is the correct output in this case? Should the default linewidth ignore transforms (e.g. the linewidth should default to 1 * fabs(descriminant of current transform)?

not sure what the answer is, the only thing that is clear is that the Inkscape result is not consistent with Adobe Reader. Adobe seems to produce a default linewidth that is visually 1 px, regardless of whatever transforms may be present.

I checked the PDF. This behavior is resolution-dependent, so it can't be replicated exactly. As an approximation, we can convert such lines to 1px strokes.

See also: bug #448286

Partial fix (hairline strokes rendered as 1px lines compensated for transform) in trunk r13143.
Backported to stable in r10016.