The Apache EBCDIC Port
Warning: This document
has not been updated to take into account changes made in
the 2.0 version of the Apache HTTP Server. Some of the
information may still be relevant, but please use it with care.
Version 1.3 of the Apache HTTP Server is the first version
which includes a port to a (non-ASCII) mainframe machine which
uses the EBCDIC character set as its native codeset.
(It is the SIEMENS family of mainframes running the BS2000/OSD
operating system. This mainframe OS nowadays features a
SVR4-derived POSIX subsystem).
The port was started initially to
- prove the feasibility of porting the Apache HTTP server to
- find a "worthy and capable" successor for the venerable
(which was ported a couple of years ago), and to
- prove that Apache's preforking process model can on this
platform easily outperform the accept-fork-serve model used
by CERN by a factor of 5 or more.
This document serves as a rationale to describe some of the
design decisions of the port to this machine.
One objective of the EBCDIC port was to maintain enough
backwards compatibility with the (EBCDIC) CERN server to make
the transition to the new server attractive and easy. This
required the addition of a configurable method to define
whether a HTML document was stored in ASCII (the only format
accepted by the old server) or in EBCDIC (the native document
format in the POSIX subsystem, and therefore the only realistic
format in which the other POSIX tools like
sed could operate on the documents). The current
solution to this is a "pseudo-MIME-format" which is intercepted
and interpreted by the Apache server (see below). Future versions
might solve the problem by defining an "ebcdic-handler" for all
documents which must be converted.
Since all Apache input and output is based upon the BUFF
data type and its methods, the easiest solution was to add the
conversion to the BUFF handling routines. The conversion must
be settable at any time, so a BUFF flag was added which defines
whether a BUFF object has currently enabled conversion or not.
This flag is modified at several points in the HTTP
- set before a request is received
(because the request and the request header lines are always
in ASCII format)
- set/unset when the request body is
received - depending on the content type of the request body
(because the request body may contain ASCII text or a binary
- set before a reply header is sent
(because the response header lines are always in ASCII
- set/unset when the response body is sent
- depending on the content type of the response body (because
the response body may contain text or a binary file)
The relevant changes in the source are
into two categories:
Code which is needed for any EBCDIC based machine.
This includes character translations, differences in
contiguity of the two character sets, flags which
indicate which part of the HTTP protocol has to be
converted and which part doesn't etc.
Code which is needed for the SIEMENS BS2000/OSD
mainframe platform only. This deals with include file
differences and socket implementation topics which are
only required on the BS2000/OSD platform.
The possibility to translate between ASCII and EBCDIC at
the socket level (on BS2000 POSIX, there is a socket option
which supports this) was intentionally not chosen,
because the byte stream at the HTTP protocol level consists
of a mixture of protocol related strings and non-protocol
related raw file data. HTTP protocol strings are always
encoded in ASCII (the
GET request, any Header: lines,
the chunking information etc.) whereas the file transfer
parts (i.e., GIF images, CGI output etc.)
should usually be just "passed through" by the server. This
separation between "protocol string" and "raw data" is
reflected in the server code by functions like
rvputs() for strings, and functions like
bwrite() for binary data. A global translation
of everything would therefore be inadequate.
(In the case of text files of course, provisions must be
made so that EBCDIC documents are always served in
This port therefore features a built-in protocol level
conversion for the server-internal strings (which the
compiler translated to EBCDIC strings) and thus for all
server-generated documents. The hard coded ASCII escapes
\015 which are ubiquitous
in the server code are an exception: they are already the binary
encoding of the ASCII
must not be converted to ASCII a second time.
This exception is only relevant for server-generated strings;
and external EBCDIC documents are not expected to
contain ASCII newline characters.
By examining the call hierarchy for the BUFF management
routines, I added an "ebcdic/ascii conversion layer" which
would be crossed on every puts/write/get/gets, and a
conversion flag which allowed enabling/disabling the
conversions on-the-fly. Usually, a document crosses this
layer twice from its origin source (a file or CGI output) to
its destination (the requesting client):
Apache -> client.
The server can now read the header lines of a CGI-script
output in EBCDIC format, and then find out that the remainder
of the script's output is in ASCII (like in the case of the
output of a WWW Counter program: the document body contains a
GIF image). All header processing is done in the native
EBCDIC format; the server then determines, based on the type
of document being served, whether the document body (except
for the chunking information, of course) is in ASCII already
or must be converted from EBCDIC.
For Text documents (MIME types text/plain, text/html
etc.), an implicit translation to ASCII can be
used, or (if the users prefer to store some documents in
raw ASCII form for faster serving, or because the files
reside on a NFS-mounted directory tree) can be served
to serve files with the suffix
.ahtml as a
text/html document without implicit
conversion (and suffix
.ascii as ASCII
text/plain), use the directives:
AddType text/x-ascii-html .ahtml
AddType text/x-ascii-plain .ascii
text/foo MIME type can be
served as "raw ASCII" by configuring a MIME type
text/x-ascii-foo" for it using
Non-text documents are always served "binary" without
conversion. This seems to be the most sensible choice for,
.e.g., GIF/ZIP/AU file types. This of course
requires the user to copy them to the mainframe host using
rcp -b" binary switch.
Server parsed files are always assumed to be in native
(i.e., EBCDIC) format as used on the machine, and
are converted after processing.
For CGI output, the CGI script determines whether a
conversion is needed or not: by setting the appropriate
Content-Type, text files can be converted, or GIF output can
be passed through unmodified. An example for the latter case
is the wwwcount program which we ported as well.
All files with a
Content-Type: which does not
text/ are regarded as binary
files by the server and are not subject to any conversion.
Examples for binary files are GIF images, gzip-compressed files
and the like.
When exchanging binary files between the mainframe host and
a Unix machine or Windows PC, be sure to use the ftp "binary"
TYPE I) command, or use the
rcp -b command from the mainframe host (the
-b switch is not supported in unix
The default assumption of the server is that Text Files
(i.e., all files whose
text/) are stored in the native
character set of the host, EBCDIC.
SSI documents must currently be stored in EBCDIC only.
No provision is made to convert it from ASCII before