CINXE.COM

<h1>Common Mistakes that cause Automated Processing to Fail</h1> <p>NEWS 2017-02-09: We have upgraded our TeX system to <a href="http://tug.org/texlive">TeXLive 2016</a> and revised the macro package tree for new submissions. All updates available as of 2016-10-30 have been included in our installation of the texlive system. </p> <p>NEWS 2011-12-06: We have upgraded our TeX system to <a href="http://tug.org/texlive/">texlive 2011</a> and revised the macro package tree for new submissions. All updates available as of 12/06/2011 have been included in our installation of the texlive system. </p> <p>NEWS 2009-12-31: We have upgraded our TeX system to <a href="http://tug.org/texlive/">texlive 2009</a> and revised the macro package tree for new submissions. All updates available as of 12/31/2009 have been included in our installation of the texlive system. See <a href="texlive">texlive 2009 transition help</a> for possible problems that may result from this upgrade.</p> <p>2006-11-02: We have upgraded our TeX system to teTeX3 and revised the macro package tree for new submissions. See <a href="tetex3">teTeX3 transition help</a> for possible problems that may result from this upgrade.</p> <p> Look through these common mistakes if your TeX/LaTeX submission failed: </p> <ul> <li><a href="#abs_filenames">Absolute filenames</a></li> <li><a href="#case_filenames">Upper-case vs Lower-case filenames</a></li> <li><a href="#missing_macro">Missing style/macro files</a></li> <li><a href="#untested">Last minute untested changes</a></li> <li><a href="#intervention">User intervention</a></li> <li><a href="#concatenated">Multiple files concatenated</a></li> <li><a href="#old_style">Use of old or non-standard style files</a></li> <li><a href="#dvips_clash">Style files incompatible with dvips</a></li> <li><a href="#other_formats">Sources that rely on something other than TeX or LaTeX</a></li> <li><a href="#fonts">Use of unusual/uncommon fonts</a></li> <li><a href="#protect">Unprotected <code>\cite</code> and other macros inside figure <code>\caption</code> commands, "<code>line too long</code>" error</a></li> <li><a href="#auto_ignore">Marking files to be ignored</a></li> <li><a href="#bbox">"<code>! Missing number, treated as zero</code>" error</a></li> <li><a href="#nohypertex">To disable HyperTeX</a></li> <li><a href="#include_subdir">"<code>Can't write subdir/file.aux</code>" and other problems with write permissions during TeX processing</a></li> <li><a href="#double_subscript">"<code>! Double subscript/superscript</code>" errors</a></li> <li><a href="#bad_pdfmark">PDF conversion failure in papers with complex section structure</a></li> <li><a href="#ifpdf">arXiv system attempts processing with PDFLaTeX for submissions which are regular latex</a></li> <li><a href="feynmf#overwrite">Why does my submission fail the automatic TeXing procedure when I use Feynmf?</a></li> <li><a href="#wrongtex">Why does my submission fail to recognize the main tex file?</a></li> <li><a href="#psbad">Problems with inclusion of binary or other bitmap figures; <code>PS BAD</code> warnings</a></li> </ul> <hr noshade="noshade" /> <br/> <dl> <dt><a name="abs_filenames"></a><b>Absolute filenames</b></dt> <dd> <p> When including figures/style/class/macro files, you must use relative filenames instead of absolute filenames. For example, <object><pre> myfigure.png </pre></object> is correct, while: <object><pre> /users/staff/fred/myarticle/myfigure.png </pre></object> is not. </p> <p> Absolute filenames make it impossible for anyone to use the source without modifying it. Unless your TeX system happens to have the same directory structure as ours, then our automated processing will fail. </p> <br /></dd> <dt><a name="case_filenames"></a><b>Upper-case vs Lower-case filenames</b></dt> <dd> <p> If you use a computer with case-insensitive filenames (e.g., Windows), be sure that the case of any filenames referred to in your TeX file matches exactly the files that are uploaded. For example, the command: <object><pre> \includegraphics{figure1.eps} </pre></object> will not work if the file is uploaded as <object><pre> FIGURE1.EPS </pre></object> </p> <p> because filenames are case-sensitive on our system. </p> <br /></dd> <dt><a name="missing_macro"></a><b>Missing style/macro files</b></dt> <dd> <p> Some authors write their own style/macro files (or modify standard ones), but forget to include them with the source. Be sure to include with the source any style/macro files that we don't have.</p> <p><b>Note:</b> non-standard style/macro files provided along with the source must <b>NOT be attached</b> to a paper. Instead, submit them as separate files, or use <b>tar</b> or <b>zip</b> to combine them with the source files before submitting.</p> <br /></dd> <dt><a name="untested"></a><b>Last minute untested changes</b></dt> <dd><p>All too often, an author will make last minute changes to the source, but won't test it to see if the results will TeX correctly. You can save yourself a lot of trouble by testing that everything works properly before submitting your paper.</p></dd> <dt><a name="intervention"></a><b>User intervention</b></dt> <dd><p>Our source to postscript conversion system (called "AutoTeX") is <i>fully automated</i>. There is no genie present to answer questions such as: </p> <pre> Would you like (P)ortrait or (L)andscape ? - answer P/L </pre> <p>If your source needs to ask these types of question, please create a file called <code>filename.inp</code> (where <code>filename</code> matches the file it is to act on) that contains suitable responses. </p> <p>Your <code>.inp</code> file can be first tested on your local machine by running the following command:</p> <pre> $ cat filename.inp | tex filename.tex </pre> <br /></dd> <dt><a name="concatenated"></a> <b>Multiple files concatenated</b></dt> <dd><p>Do not submit several files all concatenated into one file, such as the Scientific Workplace's <code>.rap</code> file type -- such files cannot be separated automatically and may fail to process in unexpected ways. Instead, please create a <code>.tar.gz</code> or <code>.zip</code> file. </p> <br /></dd> <dt><a name="old_style"></a> <b>Use of old or non-standard style files</b></dt> <dd><p>Style files change with time, and you (or we) might be using old versions of style files. We try to keep up to date, so if problems arise due to style files, please check that the version you are using is current.</p> <p> Some well-known style files, e.g., <code>epsf.sty</code> and <code>epsfig.sty</code>, have been altered by TeX programmers, then uploaded to web sites under the same filename. This is <b>very bad practice</b> and anti-social behavior, since we end up with multiple versions of <i>standard</i> styles. arXiv will make every effort to support the most recent <b>and</b> official versions of standard style files. </p> <p> As an example, the <code>graphics</code> and <code>graphicx</code> packages have been standard since the release of LaTeX 2e and are the recommended graphics inclusion macros. These packages provide the most portable and reliable method of including graphics. </p> </dd> <dt><a name="dvips_clash"></a> <b>Style files incompatible with dvips</b></dt> <dd><p>Style files that are incompatible with <code>dvips</code>, e.g., <code>epsbox.sty</code>, can cause problems. If the DVI requires exotic versions of <code>dvips</code> (e.g., <code>jdvi2kps</code>) to produce postscript, then our automated system will <b>fail</b>. </p> <p>Please make every effort to avoid using non-standard styles and <code>dvi2ps</code> utilities. Few people will be able to process papers in these non-standard formats.</p> <br /></dd> <dt><a name="other_formats"></a> <b>Sources that rely on something other than TeX or LaTeX</b></dt> <dd><p>All papers should be formatted so they work with TeX or LaTeX (as appropriate). If you use other formats (as opposed to style files or macros), e.g., <b>AMS-TeX</b>, <b>AMS-LaTeX</b>, or <b>aatex</b>, then please add the appropriate line to the top of the source, e.g.:</p> <pre> %&amp;amslplain </pre> for <b>AMS-LaTeX</b> (based on amslatex version 1.1), <pre> \input amstex </pre> or <pre> %&amp;amstex </pre> <p>for <b>AMS-TeX</b>, and</p> <pre> \input cp-aa.tex </pre> or <pre> %&amp;cp-aa </pre> <p> for Springer-Verlag's Plain TeX Astronomy &amp; Astrophysics macros, <b>cp-aa</b> (also known as <b>aatex</b>).</p> <p>Anything that relies on something <b>other than TeX or (PDF)LaTeX</b> will fail. At this time arXiv does not support processing with: XeTeX and its variants including LuaTeX, LyX, or PDFTeX.</p> <br /></dd> <dt><a name="fonts"></a> <b>Use of unusual/uncommon fonts</b></dt> <dd><p>Authors should keep in mind that if they use unusual fonts, many potential readers of their work won't have them installed.</p> <p>Because it would require significant admin time to install and maintain non-standard fonts, we do not generally support such fonts.</p> <p> It is, however, possible to include your local metafont <code>.mf</code> files as well as your local <code>fontmap.map</code> file. This will require special handling with a <code><a href="/help/00README#fontmap">00README.XXX</a></code> file with the appropriate <b>fontmap</b> directive. </p> <br /></dd> <dt><a name="protect"></a> <b>Unprotected <code>\cite</code> and other macros inside figure <code>\caption</code> commands, "<code>line too long</code>" error</b></dt> <dd> <p>In LaTeX, any citations inside a figure <code>\caption</code> should be protected using the <code>\protect</code> command; e.g., <code>\caption{Electron spectral function from \protect\cite{spectral}}</code>. This delays the expansion of the citation until the second latex pass, i.e., when the reference has been defined. </p> <br /></dd> <dt><a name="auto_ignore"></a> <b>Marking files to be ignored</b></dt> <dd> <p>If you need to include files in the source that should not be processed by the automated system, you may do either of the following: </p> <ol> <li>Add: <pre> %auto-ignore </pre> <p>near the top of the file. This directive can be anywhere in the first 10 lines of the file, and anywhere on the line. It should appear before any TeX or LaTeX commands, since otherwise they would be recognized first. For example:</p> <pre> %auto-ignore This is a README file for paper hep-ex/9901003 More data for our experiment is available at http://www.some.where/else </pre></li> <li>Include a file <code><a href="/help/00README">00README.XXX</a></code> with your submission that includes the line: <pre> filename ignore </pre> for each file that should be ignored.</li> </ol> <br /></dd> <dt><a name="bbox"></a> <b>"<code>! Missing number, treated as zero</code>" error</b></dt> <dd><p> If you use <code>epsf</code> to include PostScript figures you must make sure that the <code>%%BoundingBox</code> definitions are near the start of the PostScript figure files. In order to reduce processing time, our TeX system does not scan the whole of each included file. </p> <p> If you have a file with the following structure: </p> <pre> %!PS-Adobe-3.0 ... ... %%BoundingBox: (atend) ... [ bulk of PS file in here ] ... %%BoundingBox: 0 10 234 456 ... %%EOF </pre> then simply move the <code>%%BoundingBox</code> line to the top: <pre> %!PS-Adobe-3.0 ... ... %%BoundingBox: 0 10 234 456 ... [ bulk of PS file in here ] ... %%EOF </pre> <br /></dd> <dt><a name="nohypertex"></a> <b>To disable HyperTeX</b></dt> <dd><p> By default, our TeX system uses <a href="/hypertex">HyperTeX</a> to add hyperlinks between references, sections and equations within your paper. These show up in the PDF (and in the PostScript with some viewers). </p> <p> HyperTeX conflicts with a few style and class files. If you think this is a problem, you can disable HyperTeX for your submission by including a file <code><a href="/help/00README">00README.XXX</a></code>. It should contain the line: </p> <pre> nohypertex </pre> <p> Note that HyperTeX changes the way citations appear in some styles -- ranges will be represented as [11,&nbsp;12,&nbsp;13] instead of [11-13]. This is necessary for HyperTeX to be able to make individual links to each citation. Unless you feel very strongly about this we recommend leaving HyperTeX on. </p></dd> <dt><a name="include_subdir"></a> <b>"<code>Can't write subdir/file.aux</code>" and other problems with write permissions during TeX processing</b></dt> <dd> <p> In our system, only the top level directory is granted write permission during processing. Attempts to write files to subdirectories will fail. </p> <p> All files included via <code>\include</code> instead of <code>\input</code> must be in the top level directory. This is because the <code>\include</code> command attempts to write a separate <code>.aux</code> file in the same directory as the included file. For example: </p> <pre> \input{file} %OK, does not create separate .aux file \input{subdir/file} %OK, does not create separate .aux file \include{file} %OK because file.aux can be written \include{subdir/file} %WILL FAIL fail because sub/file.aux cannot be written </pre> <br /></dd> <dt><a name="double_subscript"></a> <b>"<code>! Double subscript/superscript</code>" errors</b></dt> <dd> <p>Our TeX system complains about double subscripts (and superscripts) because <code>a_x_y</code> could be read as <code>a_{x_y}</code> or <code>{a_x}_y</code> or <code>a_x{}_y</code> or even <code>a_{xy}</code>. These are not the same since the character size and position is affected.</p> <p>Some older TeX systems would automatically substitute <code>a_x{}_y</code> without the need for user intervention, but current TeX systems (including ours) will not do this. As a result, <b>the appropriate interpretation must be explicitly specified</b>.</p> <p>If you are curious about the differences, see <a href="doublesubscript">these examples</a>.</p> <br /> </dd> <dt><a name="bad_pdfmark"></a> <b>PDF conversion failure in papers with complex section structure</b></dt> <dd> <p>In some papers with &quot;elaborate&quot; section structure, hyperref can generate PostScript that contains bad pdfmarks which break PDF conversion. This problem affects mostly longer papers, reviews, theses, etc., i.e. those papers where a linked table of contents (TOC) and document outline would be most useful!</p> <p>The reason is that the section counter (used for the names of destination links) is reset by authors and certain macros to control the numbering of appendix sections. The result is failure of ps2pdf conversion due to ambiguous pdfmarks or conflicting subsection count in the pdfmarks.</p> <p>The simple cure is to <a href="#nohypertex">disable HyperTex</a> but a better work-around is to simply switch off bookmarks (i.e. document outline) while leaving normal document linking on. You can do this by adding the following line to the preamble:</p> <pre> \usepackage[bookmarks=false]{hyperref} </pre> <br /> </dd> <dt><a name="ifpdf"></a> <b>arXiv system attempts processing with PDFLaTeX for submissions which are regular latex</b></dt> <dd> <p>A common mistake made by authors as well as many macro packages is incorrect testing for \pdfoutput to decide whether pdflatex is run in dvi mode or pdf mode, or whether the processing is done in regular latex mode. The underlying engines used to be different and a simple test for \ifx\pdfoutput\undefined was sufficient to distinguish between all options. This is no longer the case, because the underlying engine is the same for all 3 cases and therefore the value of the \pdfoutput parameter has to be tested, too.</p> <p>That is, a common (but incorrect) testing sequence might look like: "\ifx\pdfoutput\undefined .... \else ..... \fi". Symptoms of this mistake might be:</p> <ul> <li>error messages "Option clash for package ...."</li> <li>some unexpected message about various pdf operators</li> <li>failing figure inclusions (due to unrecognized extension)</li> </ul> <p>The most reliable way to accomplish conditional branching in the TeX source where necessary is instead via the ifpdf package:</p> <pre> \usepackage{ifpdf} \ifpdf do something pdflatex specific here \else do something for regular latex or pdflatex in dvi mode \fi </pre> <p>Note that the graphics package and the hyperref package are smart enough to figure this out on their own. You do <b>not</b> need to specify any driver for these packages.</p> <p>For more information, see:</p> <a href="https://www.ctan.org/pkg/ifpdf">https://www.ctan.org/pkg/ifpdf</a> <p>You can download the ifpdf macro from the above link as well, if it is not already a part of your tex tree.</p> <p> In the extremely rare event that your submision still does not correctly identify itself as latex, and you are absolutely sure, you can add the line: <pre> %&latex </pre> as the very first line of your tex file. </p> <br /> </dd> <dt><a name="wrongtex"></a> <b>Why does arXiv's system fail to recognize the main tex file?</b> </dt> <dd> <p> It is possible in writing your latex code to include your <code>\documentclass</code> directive in a file other than the main .tex file. While this is perfectly reasonable for a human who's compling to know which of the tex files is the main one (even when using something obvious as the filename, such as <code>ms.tex</code>), our AutoTeX system will attempt to process whichever file has the <code>\documentclass</code> directive as the main tex file. </p> <p> Note that the system does not process using <code>Makefile</code> or any other manifest-type files. </p> <br /> </dd> </dl> <dt><a name="psbad"></a> <b>Problems with inclusion of binary or other bitmap figures; <code>PS BAD</code> warnings</b></dt> <dd> <p><em>Update 2011-12-06: arXiv's default dvips configuration was changed to retain comments. The following does not apply to papers received since then.</em></p> <p> By default, our TeX system tells <code>dvips</code> to strip comment lines from included PostScript figures. This is usually the correct thing to do because it prevents DSC (Document Structuring Comments) lines from being included from figures. If included, these would likely break the DSC structure of the final PostScript (the DSC structure is what allows viewers to display an index of page numbers). </p> <p> Lines starting with '<code>%</code>' in the included PostScript files are identified as comments, but sometimes PostScript figures include blocks of data with lines starting with '<code>%</code>' that are not comments. Removal of these lines may break the final PostScript. You can turn off the removal of comments from included figures for your submission by including a file <code><a href="/help/00README">00README.XXX</a></code>. It should contain an instruction specific to the filename of the dvi file: </p> <pre> filename.dvi keepcomments </pre> <p> You can test the effect of this yourself by comparing the output of </p> <pre> $ dvips -R <em>-K1</em> yourfile.dvi -o DSCstripped.ps $ dvips -R <em>-K0</em> yourfile.dvi -o DSCkept.ps </pre> <p> Note that this whole issue arises because Adobe decided to use the ignored &quot;Comments&quot; from the PostScript standard to provide additional structure to regular PostScript files, which leads to complications for programs that rely on proper DSC structure when two or more such files are included in each other. </p> <br /></dd>

Pages: 1 2 3 4 5 6 7 8 9 10