How many publications do I have?
Actually, I don’t care too much about that; but if I were younger and applying for promotion, it would actually matter. But suppose you wanted to answer the question, or indeed to get copies or reviews of the papers, or to find citation counts.
You could turn to the abstracting/reviewing websites, MathSciNet or Zentralblatt Math. They list, respectively, 320 and 328 publications, Zentralblatt giving the additional information that this includes 16 books. MathSciNet also finds 3252 citations of my work. Of course, both of these sites give access to reviews of the papers, and in some cases it is possible to get the published papers from there. But this depends on having the relevant subscriptions (as indeed does access to these two sites).
What about Web of Science? Well, I completely failed to find how to search for just my publications, and there are lots of people called P Cameron beavering away! Accurate author identification really matters.
Lots of people use Google Scholar, and it is not hard to see why. This site credits me with 381 publications, having 11351 citations. How do they get so many? I went down to the end of the list and found all sorts of things, some just about legitimate (documents put on the web but never published), some just crazy (fififiifiifli%fifi§§%fi._, by P. J. Cameron and N. M. Singhi, anyone?) There are ways of tidying this up, but it hardly seems worth the trouble. As to finding three times as many citations as MathSciNet does, I really don’t know how they manage this.
ORCiD tries to be accurate and reliable; it links to other databases and also allows me to edit entries easily. It lists 265 publications. Some of these were imported from Scopus, which lists 249 documents and 2991 citations. Also, increasingly, publishers are allowing you to log in via ORCiD rather than needing a separate login for each journal.
ORCiD gives DOIs for the papers it lists, which means that you can get anything (even book chapters) if you have the appropriate subscription (usually the case if you are logged in to a university account, though this may change).
Scopus is a commercial site; the information there seems fairly reliable (if incomplete), with one exception. Here is the list of my interests on Scopus: Mathematics, Computer Science, Physics and Astronomy, Social Sciences, Biochemistry, Genetics and Molecular Biology, Engineering, Decision Sciences, Psychology. Most of these are spurious. (I do have one paper in each of the Communications in Mathematical Physics and the Journal of Mathematical Psychology, but that doesn’t mean that either physics or psychology is a research interest of mine.)
DBLP gives me 128 publications of which 121 are journal articles (the others are parts in collections or informal publications). It links to everything and gives easy access to the other sources I have mentioned here. Why relatively few? It is a computer science database; frankly, I am surprised that I have as many as 128 publications that can be classified as computer science. (But I was astonished at the size of the DBLP database: they have over 8000 authors listed with the surname Wu, for example.)
I have 65 papers on the arXiv classified as mathematics, 5 computer science and 4 statistics (but there are some overlaps). Of course, I have not been putting papers on the arXiv for my entire career; first it didn’t exist, and then when it did I was too incompetent to use it, so my earliest papers there were posted by coauthors. And of course the great advantage of the arXiv is that you may freely get copies of the papers there. Most new papers of mine go there now, as well as some historical documents dug out of the files which may be of some interest.
The St Andrews repository lists 78 of my publications. These are the ones since the last-but-one research assessment (we were instructed some years ago to enter all these, and I never got round to putting in earlier ones: the process is far from straightforward). It is a repository: as well as titles, open-access copies or post-acceptance manuscripts available there.
According to my own list I have 344 publications, of which 17 are books. This is not completely accurate either. When I drew up this list some years ago, I made a few mistakes, so that one number in the range has no corresponding publication, while others have several (e.g. several chapters in the same book). Also, this list includes papers which are in press or have been accepted for publication, which it would not be reasonable for MathSciNet or Zentralblatt to know about.
So clearly MathSciNet and Zentralblatt are the most accurate sources. But if I were applying for promotion, you can see how Google Scholar would be tempting. Also, in my view, the arXiv is the most important repository.