GoogleCodeExporter
commented
9 years ago I've noted both on irc and on the forums that the next major release will have
ram feature, but it's a very difficult one to implement and experience has
shown me that such features usually take about 2 to 4 weeks to get stable
depending on the range of user data sets, and the quality of the data, which in
this case is very low, which means lots of filtering.
I had hoped for some time that /sys would grow to contain ram data, I actually
do not understand why it doesn't, but it hasn't done so at this point, so I'm
giving up on that idea and going back to the fallback of using dmidecode to
grab that data, which has several major negatives, one being that you must be
root to run dmidecode, the other being that parsing /sys is VERY fast, and much
more reliable.
But since ram is not appearing in /sys, at least not last I checked, with one
exception eec ram types, which are almost useless for most users, I decided a
while ago to just use dmidecode, which also has the advantage of giving
automatic bsd support as well without further hacks.
I have a crude outline of what -R or -M, probably -M for memory, will show,
it's a lot of data, it will show the number of memory bank arrays, usually one
or two for consumer systems, much more for servers, and I've still not seen an
actual server dmidecode output which I will need to see before planning this,
the amount of ram that each bank or array supports, per slot maxes, then the
population of the arrays, etc. The amount of information available to the
system is very finite however, sometimes there is no product name, no speed,
etc, but the capacities are generally present.
The more of these data sets: inxi -xx@14
I can get, that's the full data dump for the system, run as root, with
dmidecode installed, the better. Particularly from servers with far more ram
installed than normal, I need to build in that support from the ground up or it
will be too hard to modify smaller systems to support larger ones.
I don't have 2 to 4 free weeks at the moment nor do I expect to have that for a
while, but I can collect data in the meantime, usually the more data I have the
better, but most of inxi's datasets are generated without dmidecode installed
and not as root, so I have very few dmidecode datasets, I need at least 10 from
systems of varying age, from old redhats to bsds to real servers with tons of
ram to normal systems to laptops etc. Experience has shown me with linux/bsd
features that expecting consistency of any type in output is a major error, so
I expect inconsistency now, and try to build in support for that from the
ground up.
That's why I added the recent -w/-W weather option instead, that was a fairly
quick and easy hack that only took a few days of coding and debugging to get
stable, now stable in 1.9.7 for the time being.
memory support will be in either 1.10 or 2.0, depending on the version
numbering selected, probably 1.10 to avoid the now disgustingly popular version
number inflations made popular by firefox and google chrome browser.
It's alwasy important, of course, to grasp that inxi won't add this feature, I
will, and that takes a lot of work and time. I used to make the mistake of
thinking new features would be easy (-R raid and -s sensors come to mind as
some of my more radically wrong guesses in terms of how long it would take to
get them stable, and how many user data sets would be required to actually get
the programming to handle most cases out there that exist), but I have learned
my lesson, difficult features are not easy and always cost me roughly one month
of time as a dev, between data collection, analysis, programming, debugging,
rewriting, debugging, data collection, analysis, etc...
So yes it's the next major feature planned, but don't hold your breath for the
release day, it will come when it comes.Original comment by inxi-...@techpatterns.com on 28 May 2013 at 7:14
- Changed state: Accepted
- Added labels: Milestone-Release1.0
- Removed labels: Priority-Medium, Type-Defect
Original issue reported on code.google.com by
leafonsw...@gmail.comon 28 May 2013 at 7:06