On Sat, Mar 15, 2025 at 12:33:40PM -0500, Dima Pasechnik wrote:
> On 15 March 2025 07:32:08 GMT-05:00, Qian Yun <oldk1...@gmail.com> wrote:
> >Hi Dima,
> >
> >About the FriCAS-SageMath interface, and about making
> >FriCAS a standard package in SageMath, could you open
> >a meta bug in GitHub to track the progress?
> >
> >I guess a better interface (C-based or socket-based,
> >more robust and performant than current text-based)
> >is needed. And SageMath needs to support "RootSum",
> >or the interface needs another way to translate
> >the "%%A1" algebraic numbers.
> >
> >
> >Hi Waldek,
> >
> >One of the FriCAS development goals is:
> >"make it easier for external programs to interface with FriCAS"
> >
> >Do you think communication via socket is a good idea,
>
> Currently SageMath communicates with FriCAS via a socket interface (pexpect).
> No files are written or read.
The big deal is about converting representations. Parsing/unparsing
takes time and is tricky in many cases. Machine oriented representation
should be much faster to transmit/convert.
When taking about speed one aspect is latency. Socket communication
involves system calls and unfortunately few years ago due to spectre
fixes effective latency of system call got significantly higher.
Function calls within the same process are much cheaper, but calls
between significantly different languages tends to be much more
expensive than calls within a single language. Another aspect
is throughput, that is speed of transmitting data. Here cost is
mainly due to volume and needed transformations. In particular
translation to textual forms tend to be more expensive.
IIUC pexpect works at tty level, which is different from sockets.
Tty handling may significantly increase cost compared to files
(I do not know if this is significant factor in FriCAS interface).
--
Waldek Hebisch
--
You received this message because you are subscribed to the Google Groups
"FriCAS - computer algebra system" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to fricas-devel+unsubscr...@googlegroups.com.
To view this discussion visit
https://groups.google.com/d/msgid/fricas-devel/Z9XBTdbcCKVq0BI3%40fricas.org.
