I am in the process of typesetting a fairly large book using LaTeX. It usually is a straightforward task, and often it even poses interesting challenges. The best part of it is that LaTeX is a workhorse of a typesetting engine, and together with TeX it would chew enormous chunks of text without even giving a hiccup. The stability of the whole system was impressive.
Or so I thought…
Last Friday I emailed home the LaTeX code I was working on. At the time the work was about 20% of the final book, and used only three floats, although a significant number of packages were loaded.
I began to work on it at home, using the code I emailed myself from office. I generated each now and then some PDFs, while continuing my typesetting.
Suddenly, I noticed that I’m not seeing my PDF file. Moreover, it seemed that pdfLaTeX was stuck near the end of the document, inactive, producing no output, and eating ~95% of CPU.
Oh my… I couldn’t do anything save for killing the process. Trying to compile the source with a different engine, such as latex or pslatex didn’t do anything. Each of them was stuck eating way too much CPU at precisely the same point.
I hunted the probable cause of the problem, and it looked like a certain float located near the end of the document was the culprit. I commented it out (for LaTeX, the percentile sign, % is the comment marker) and voilá! the document compiled like a charm. I was dumbfounded to say the least.
Perhaps a filesystem problem? I did fsck three times with badblocks and optimize directories, and nothing came out. Phew, this was not the cause, thankfully.
At that moment it was near midnight and I was so sleepy I couldn’t think with clarity. So, I hit the sheets and slept soundly till Saturday… noon… *blushes* ;). Then I woke up, and after getting up to speed, booted the laptop, decommented the suspect code, and lo and behold, the PDF came out like business as usual.
I’m still amazed and dumbfounded, and at a loss for an answer. The most probable explanation was that the laptop’s refrigeration system couldn’t cool either the CPU or some other component, and the instability was because of the overheating. Any other explanation…?