Basically, I just added three new process properties (io_read, io_write, io_perc - representing
the amount of I/O done by the process during the update interval) and $top_io, that sorts
processes based on io_perc.
Atm, it's completely #ifdef'd, since it requires kernel support. But that creates some wierd
looking syntax at some places, so it may be better to remove some ifdefs. It even may be
possible to completely remove the ifdefs (ie. convert them to #ifdef linux) since the code will
compile just fine even if the kernel doesn't support I/O accounting. I'll leave that for someone
else to decide.
When a '#' is found in TEXT, section remove_comments() is called. However, that function doesn't
limit itself to removing that particular comment. Instead, it greedily removes all "comments" from
the string, including those in ${color $abcdef} constructs. This, obviously, makes a mess of
things.
Solution: I added a new function, remove_comment(), that only removes the current comment.
Note: Atm, the function doesn't delete the newline, since that was the current behavior, and is
the natural one for the config section. However, for the TEXT it seems better to delete the \n
too, to avoid blank lines.
Note2: this also fixes the bug in strfold(), which incorectly copied the string, truncating it by
one byte.
Signed-off-by: Brenden Matthews <brenden@rty.ca>
Closing the X display on reload caused some weirdness. We can just
leave it open between reloads. For whatever reason we had a macro
inside quotes for the default net device.
Changed llua_getinteger() to llua_getnumber() returning a double, so
that you can use floating point values in graphs etc. Lua graph will
scale like other graphs (except execgraph) by default now, and you can
manually set a scale with the scale argument.
In general, initialising last_update_time to next_update_time upon
startup doesn't make sense, as some update functions check for the
distance between the two being higher than a given epsilon to prevent
updating too often. This means that they won't trigger when
update_stuff() is being run for the first time.
When you started "$updates|${addr eth0}" displayed "0|No Address",
after the first update you got something normal like "1|192.168.0.1"
This patch should fix this.
part of the template-handling code was in #ifdef X11, which caused conky to dump core when
compiled without x11.
as a bonus, I changed tests !=' ' to !isspace in two places as it was causing strange effects
Signed-off-by: Brenden Matthews <brenden@rty.ca>
Conky would dump core when encountering templates with no parameters (both $templateX and
${templateX}) and when the line contained improperly nested {}.
Signed-off-by: Brenden Matthews <brenden@rty.ca>
The if_existing with 2 arguments checking whether a file contains a specified
string did not work. This is fixed by this patch.
Signed-off-by: Brenden Matthews <brenden@rty.ca>
See docs for more details on 'temperature gradients', which can be
turned on with the -t switch on graphs. Also improved parsing of
options with $execgraph and $execigraph, but may cause some breakage.
Altough the following configuration is wrong, it should only produce a error
about a missing $endif but instead it also produced a segmentation fault:
TEXT
$if_gw$gw_ip$else
All lines behind a comment line were also removed, comments should also be able
to start in the middle of line, \# shouldn't be replaced if it's in a comment
Also undid the last change because sony.h was not included, feel free to submit
it again if you include this file and other sourcefiles needed (there is also
no definition of get_sony_fanspeed). If sony systems already have this file and
a definition of this function, please update the configure file so that support
for this is disabled by default and can be enabled by sony owners.
Added the (incomplete) check_docs.py, to 'synchronize' the docs with the
code, as well as vim/nano syntax stuff.
Removed some unused OBJ_* stuff from text_objects.h, and updated docs
with some missing things. Also removed a couple deprecated objects
which were still documented.
The inconsistent naming of the cpu parameter in the docs led me to this,
so I also simplified parsing by introducing the macro SCAN_CPU(). Note
that this introduces a syntactical change to the config: the cpuN
argument now has to be passed at first position to $cpugraph.
Moved interface_up(...) from linux.{c.h} to common.{c,h} and taught it
to check for ENXIO as well to make it work on FreeBSD.
Signed-off-by: Nikos Ntarmos <ntarmos@cs.uoi.gr>
Moving sysfs-related functions and defs out of common.h and into
linux.h, as sysfs exists only on Linux, and updating openbsd.c,
netbsd.c, and freebsd.c accordingly.
Signed-off-by: Nikos Ntarmos <ntarmos@cs.uoi.gr>
Typo: __LINUX__ was used in a conditional code block, while the rest of
the code uses __linux__
data: free_text_objects(...) #define's 'data' to be 'obj->data' to make
the code more compact. The linux-specific parts of the code still
used obj->data, which should expand to obj->obj->data, which is
wrong.
Signed-off-by: Nikos Ntarmos <ntarmos@cs.uoi.gr>
This object makes use of the possibility to escape dollar signs in TEXT.
Take the following example in the conkyrc:
| $${downspeed ${gw_iface}}
will be evaluated to (assuming the gw_iface is eth0):
| ${downspeed eth0}
and finally interpreted to print the gateway interface's downspeed rate.
The idea found in the code is any double dollar ($$) is being treated as
explicit dollar sign ($) instead of the start of a text object
reference. Due to missing update of the 's' variable, when creating a
text object for the text following the second dollar sign, the later was
prepended, leading to a double dollar in the output.
Use dynamic allocated memory for skey and svalue,
allocate the length of value so that there are no limits,
but don't use the %a gnu extension to maintain compatibility
This patch is written by a anonymous author, see patch on sf.net
id #2663691 . I changed it a bit so that it can't overwrite
existing environment variables and updated the docs and syntaxfiles
As a side effect, font and colour specific stuff got outsourced, too.
This is because the apropriate functions are used by conky.c as well as
specials.c, so they should be kept on their own.
In the long term I hope for positive impact on the X11 integration mess.
(Take e.g. the mass of X11 ifdefs cluttering conky.c.) Though this
commit contains no optimisations in this direction, just plain
outsourcing, to ensure minimised (intended: none) changes to conky's
interface behaviour.
In fact these two objects share a lot of code, so the bigger plan is to
merge them into a single file to share equal code.
This should not change anything to the code flow, besides making conky
startup a bit more robust due to less use of CRIT_ERR.
Well, not really. I don't dare putting it all together into a single
source file, as that would just not make sense. Instead, this patch
eliminates the --enable-smapi configure option and all related checks,
replacing them with the equivalents from --enable-ibm.
Despite this is actually not much code and could stay inside linux.c as
well, this makes sense as I want to combine smapi and ibm-acpi
functionality. Virtually every user of a notebook made by IBM/Lenovo
will want both, so this should be a feature not a bug.
Besides making the ibm-acpi objects being built optionally, this should
not change anything to the code flow.
Also, note that free() does not zero the passed pointer. So conditional
freeing always needs to look like this:
| if (var) {
| free(var);
| var = 0;
| }
* change the height of execbar and execibar to be the same as other
bars have by default (6 pixels)
* treat mixer values as percentages as they're usually 0-100 (switch
to spaced_print() with pad_percents)
* change temp_print() from snprintf() to spaced_print()
* remove decimals from temp_print() as none of the current sources can
supply values smaller than 1 degree (C or F, doesn't matter)
* add a space between number and the unit in human_readable()
* fix number printing in human_readable()
* network $upspeed and $downspeed now use human_readable()
