[Edbrowse-dev] readline input segfault

[Edbrowse-dev] readline input segfault

[Adam Thompson]

[-- Attachment #1: Type: text/plain, Size: 672 bytes --]

Hi all,

For a while I've suspected that there's a segfault in the readline input mode of edbrowse.
For various reasons I've now got a little time to debug this,
and it looks like there's an issue between the dynamic allocation used by
readline and the static line length limit used by edbrowse.
The easiest fix would be to read everything in with readline and then only take
the amount that'll fit into the statically allocated array,
but that way we potentially silently (or not so silently) drop user input.
I'm wondering if now would be the time to switch to readline based input
throughout unless there are good reasons (Windows port?) not to do this?

Cheers,
Adam.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

[Edbrowse-dev] readline input segfault

[Karl Dahlke]

> I'm wondering if now would be the time to switch to readline based input
> throughout unless there are good reasons (Windows port?) not to do this?

Well one good reason is I don't like readline input and never use it.
I think it also crashes the windows port, as you surmised,
though that we really should fix.
And your bug I also need to look into. Thanks.

Karl Dahlke

[Edbrowse-dev] readline input segfault

[Karl Dahlke]

> For a while I've suspected that there's a segfault in the readline input mode of edbrowse.

Believe this is fixed now.
I switched to readline and typed a line much longer than 256 and all is well.

Karl Dahlke

Re: [Edbrowse-dev] readline input segfault

[Adam Thompson]

[-- Attachment #1: Type: text/plain, Size: 526 bytes --]

On Sat, Nov 14, 2015 at 09:42:16AM -0500, Karl Dahlke wrote:
> > I'm wondering if now would be the time to switch to readline based input
> > throughout unless there are good reasons (Windows port?) not to do this?
> 
> Well one good reason is I don't like readline input and never use it.

Fair enough. Out of interest, any particular reason?

> I think it also crashes the windows port, as you surmised,
> though that we really should fix.

Indeed, that is somewhat unfortunate.
Any idea why?

Cheers,
Adam.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

Re: [Edbrowse-dev] readline input segfault

[Adam Thompson]

[-- Attachment #1: Type: text/plain, Size: 468 bytes --]

On Sat, Nov 14, 2015 at 10:35:04AM -0500, Karl Dahlke wrote:
> > For a while I've suspected that there's a segfault in the readline input mode of edbrowse.
> 
> Believe this is fixed now.
> I switched to readline and typed a line much longer than 256 and all is well.

I can confirm in my testing it seems to be fixed, thanks.
Hopefully no more typing long paragraphs only to have my document (and browsing
windows) disappear in a segfault.

Cheers,
Adam.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]