I tried to compile an article, the source code of which is downloadable, to an ebook with tex4ebook. After some trivial modifications:
--- ../formal.orig.tex 2018-11-23 17:05:07.032134691 +0100
+++ formal.tex 2018-11-23 17:06:05.760497686 +0100
@@ -1,8 +1,9 @@
-\documentclass{conm-p-l}
+%\documentclass{conm-p-l}
+\documentclass{amsart}
\usepackage{%
amssymb,%
%fullpage,%
-righttag,%
+%righttag,%
pb-diagram,%
lamsarrow,%
pb-lams%
with an installation of fonts (I feel very sorry, this tex file uses a lot of very ancient stuff, usually no longer available in modern tex suites), from https://ctan.org/tex-archive/macros/lamstex/tfm?lang=en and http://jardine.math.uwo.ca/lams/
I succeed to compile it with pdflatex. However, when I use tex4ebook to compile it, with configure file collected from the post where I delete the option pic-m to reduce the size of the ebook file, I met an error while running dvisvgm
DVI error: stack not empty at end of page
looking at the error log, I also see
--- warning --- Couldn't find font `lams4.htf' (char codes: 0--122)
--- warning --- Couldn't find font `lams3.htf' (char codes: 0--123)
--- warning --- Couldn't find font `lams2.htf' (char codes: 0--122)
--- warning --- Couldn't find font `lams1.htf' (char codes: 0--125)
--- warning --- Page break within a ch map/picture
--- warning --- Improper groups in \special{t4ht+}... idv[276]
--- warning --- Improper groups in \special{t4ht+}... idv[770]
--- warning --- Improper groups in \special{t4ht+}... idv[857]
--- warning --- Improper groups in \special{t4ht+}... idv[1099]
--- warning --- Improper groups in \special{t4ht+}... idv[1124]
--- warning --- Improper groups in \special{t4ht+}... idv[1131]
--- warning --- Improper groups in \special{t4ht+}... idv[1134]
--- warning --- Improper groups in \special{t4ht+}... idv[1142]
--- warning --- Improper groups in \special{t4ht+}... idv[1238]
--- warning --- Improper groups in \special{t4ht+}... idv[1238]
--- warning --- Improper groups in \special{t4ht+}... idv[1238]
I don't know how to fix this properly, without introducing pic-m to enlarge the ebook file extensively. Thanks for suggestions.
Update: I have a new test case. The source code could be downloaded at https://arxiv.org/format/1707.01799
In order to pass to tex4ebook, I made some modifications (generated by diff):
11c11
< \usepackage{bibgerm} % Deutsches BibTex
---
> %\usepackage{bibgerm} % Deutsches BibTex
1625c1625
< is given by sending $X\in G\Sp^O$ to the orthogonal $G/H$-spectrum $\Phi^H_\mathcal U X$ whose $n$-th term is given by
---
> is given by sending $X\in G\Sp^O$ to the orthogonal $G/H$-spectrum $\Phi^H_{\mathcal U} X$ whose $n$-th term is given by
1856,1857c1856,1857
< \item Let $X\in \Cyc$ be a genuine cyclotomic spectrum whose underlying spectrum is bounded below. Then there is an equivalence of spectra $\TC^\mathrm{gen}(X)\simeq \TC(X)$.
< \item Let $X \in \Cyc_p$ be a genuine $p$-cyclotomic spectrum whose underlying spectrum is bounded below. Then there is an equivalence of spectra $\TC^\mathrm{gen}(X,p) \simeq \TC(X,p)$.
---
> \item Let $X\in \Cyc$ be a genuine cyclotomic spectrum whose underlying spectrum is bounded below. Then there is an equivalence of spectra $\TC^{\mathrm{gen}}(X)\simeq \TC(X)$.
> \item Let $X \in \Cyc_p$ be a genuine $p$-cyclotomic spectrum whose underlying spectrum is bounded below. Then there is an equivalence of spectra $\TC^{\mathrm{gen}}(X,p) \simeq \TC(X,p)$.
1869c1869
< \TC^\mathrm{gen}(X)\simeq \TC(X)\ .
---
> \TC^{\mathrm{gen}}(X)\simeq \TC(X)\ .
1871c1871
< We recall the definition of $\TC^\mathrm{gen}(X)$ in Definition \ref{def:TCpgen} and diagram \eqref{tcgen} below.
---
> We recall the definition of $\TC^{\mathrm{gen}}(X)$ in Definition \ref{def:TCpgen} and diagram \eqref{tcgen} below.
1934c1934
< Now let $X$ be a genuine $p$-cyclotomic spectrum in the sense of Definition~\ref{def:genuinepcyclo}. Let us recall the definition of $\TC^\mathrm{gen}(X,p)$ by B\"okstedt--Hsiang--Madsen, \cite{BHM}. First, $X$ has genuine $C_{p^n}$-fixed points $X^{C_{p^n}}$ for all $n\geq 0$, and there are maps $F: X^{C_{p^n}}\to X^{C_{p^{n-1}}}$ for $n\geq 1$ which are the inclusion of fixed points. Moreover, for all $n\geq 1$ there are maps $R: X^{C_{p^n}}\to X^{C_{p^{n-1}}}$, and the maps $R$ and $F$ commute (coherently). The maps $R: X^{C_{p^n}}\to X^{C_{p^{n-1}}}$ arise as the composition of the map $X^{C_{p^n}}\to (\Phi^{C_p} X)^{C_{p^{n-1}}}$ that exists for any genuine $C_{p^n}$-equivariant spectrum, and the equivalence $(\Phi^{C_p} X)^{C_{p^{n-1}}}\simeq X^{C_{p^{n-1}}}$ which comes from the genuine cyclotomic structure on $X$.
---
> Now let $X$ be a genuine $p$-cyclotomic spectrum in the sense of Definition~\ref{def:genuinepcyclo}. Let us recall the definition of $\TC^{\mathrm{gen}}(X,p)$ by B\"okstedt--Hsiang--Madsen, \cite{BHM}. First, $X$ has genuine $C_{p^n}$-fixed points $X^{C_{p^n}}$ for all $n\geq 0$, and there are maps $F: X^{C_{p^n}}\to X^{C_{p^{n-1}}}$ for $n\geq 1$ which are the inclusion of fixed points. Moreover, for all $n\geq 1$ there are maps $R: X^{C_{p^n}}\to X^{C_{p^{n-1}}}$, and the maps $R$ and $F$ commute (coherently). The maps $R: X^{C_{p^n}}\to X^{C_{p^{n-1}}}$ arise as the composition of the map $X^{C_{p^n}}\to (\Phi^{C_p} X)^{C_{p^{n-1}}}$ that exists for any genuine $C_{p^n}$-equivariant spectrum, and the equivalence $(\Phi^{C_p} X)^{C_{p^{n-1}}}\simeq X^{C_{p^{n-1}}}$ which comes from the genuine cyclotomic structure on $X$.
1936c1936
< These structures determine $\TC^\mathrm{gen}(X,p)$ as follows.
---
> These structures determine $\TC^{\mathrm{gen}}(X,p)$ as follows.
1941c1941,1942
< \TC^\mathrm{gen}(X,p) & := \mathrm{Eq}\left( \xymatrix{\TR(X,p) \ar^-{\mathrm{id}}[r]<2pt> \ar_-F[r]<-2pt> & \TR(X,p) }\right)\\
---
> \TC^{\mathrm{gen}}(X,p)
> & := \mathrm{Eq}\left( \xymatrix{\TR(X,p) \ar^-{\mathrm{id}}[r]<2pt> \ar_-F[r]<-2pt> & \TR(X,p) }\right)\\
2087c2088
< \newcommand{\Prl}{\mathcal{P}\mathrm{r}^\mathrm{L}}
---
> \newcommand{\Prl}{\mathcal{P}\mathrm{r}^{\mathrm{L}}}
4048,4050c4049,4050
< \begin{altenumerate} $ $
< \item
< For every compact space $X$ the Tate valued Frobenius (or rather its refinement as in Corollary \ref{refinement}) of $\KU^X$ is given on $\pi_{0}$ by
---
> \begin{altenumerate}
> \item For every compact space $X$ the Tate valued Frobenius (or rather its refinement as in Corollary \ref{refinement}) of $\KU^X$ is given on $\pi_{0}$ by
4057,4058c4057
< \item
< Under the identification \eqref{identification}, the Frobenius
---
> \item Under the identification \eqref{identification}, the Frobenius
4552c4551
< \item The cyclic bar construction $\B^\mathrm{cyc}M$ admits a canonical $\T$-action and a canonical $\T$-equivariant map
---
> \item The cyclic bar construction $\B^{\mathrm{cyc}}M$ admits a canonical $\T$-action and a canonical $\T$-equivariant map
4554c4553
< \psi_p: \B^\mathrm{cyc}M \to (\B^\mathrm{cyc}M)^{hC_p}
---
> \psi_p: \B^{\mathrm{cyc}}M \to (\B^{\mathrm{cyc}}M)^{hC_p}
4560,4561c4559,4560
< M \ar[d]^{\Delta}\ar[r]^{i} & \B^\mathrm{cyc}M\ar[d]^{\psi_p} \\
< (M \times \ldots \times M)^{hC_p} \ar[r] & \B^\mathrm{cyc}M^{hC_p}
---
> M \ar[d]^{\Delta}\ar[r]^{i} & \B^{\mathrm{cyc}}M\ar[d]^{\psi_p} \\
> (M \times \ldots \times M)^{hC_p} \ar[r] & \B^{\mathrm{cyc}}M^{hC_p}
4688c4687
< The upper map is given by the canonical inclusion, since in the equivalence $\B^\mathrm{cyc} \Omega Y \simeq LY$ this corresponds to the inclusion of the bottom cell of the simplicial diagram $\B^\mathrm{cyc} \Omega_\bullet Y$. The lower line is also equivalent to the inclusion $\Omega Y \to LY$ under the obvious identifications, as one sees similarly. Under these identifications the left hand map corresponds to the identity map $\Omega Y \to \Omega Y$. As a result the map $LY \to LY^{hC_p} = LY$ cannot be the trivial map which sends every map to the constant map. Thus it has to be equivalent to the identity which finishes to proof.
---
> The upper map is given by the canonical inclusion, since in the equivalence $\B^{\mathrm{cyc}} \Omega Y \simeq LY$ this corresponds to the inclusion of the bottom cell of the simplicial diagram $\B^{\mathrm{cyc}} \Omega_\bullet Y$. The lower line is also equivalent to the inclusion $\Omega Y \to LY$ under the obvious identifications, as one sees similarly. Under these identifications the left hand map corresponds to the identity map $\Omega Y \to \Omega Y$. As a result the map $LY \to LY^{hC_p} = LY$ cannot be the trivial map which sends every map to the constant map. Thus it has to be equivalent to the identity which finishes to proof.
6387c6386
< \end{document}
\ No newline at end of file
---
> \end{document}