MintPup (Trusty LTS)

[fredx181]

Sorry, Fred, forgive me... it's just my distorted Scottish sense of humour - it made me laugh.

But anyway, I'll likely look into that htop weirdness anyway since I'm interested in system's RAM (and swap) usage characteristics and having tools such as free, top, and htop that give consistent and meaningful statistics is important I feel. But where I post what I find - hmmm... somewhere else

Ok for me on XenialDog thread, I am also interested in how memory "used" is calculated nowadays.

Edit: Some observations and S.O.S. in XenialDog thread here:
http://murga-linux.com/puppy/viewtopic. ... 797#903797

Edit2:

EDIT: you are surely not suggesting that Linux Mint has patched htop to give more favourable looking RAM usage statistics?

I don't know, all I know is that in DebianDog and Mintpup I think I can trust on htop, free, lxtask or conky because they all display (almost) the same amount of memory used.

Fred

[mcewanw]

Hi Fred,

I hadn't noticed your SOS and comments and calculations until right now, but you'll find I basically agree with what you say in my own report below, which I had only just completed.

My main finding is that XenialDog actually uses less RAM on boot than MintPup (and backi's 'free' utility 'used' value is only 61252 KiBytes, which I think is the correct usage, which is quite low). My second finding (belief) is that its htop utility is all screwed up (and substituting MintPup htop in XenialDog doesn't work properly - maybe it reads wrong stats from /proc/meminfo then anyway).

I also looked at free, top, htop and /proc/meminfo values for DD jessie 64, Mintpup and XenialDog. Here is what I found:

My assumption: the utilities free, top and htop process memory values they obtain from /proc/meminfo.

In that view, the most reliable indicator of memory usage is /proc/meminfo.

1. Mintpup free utility.

NOTE WELL that the units used in this utility are KiBytes (which is 1024 Bytes).

Output provided show that the calculation being used internally by the utility is:

RAM used = total - free (note well that this is the same as MemTotal - MemFree in /proc/meminfo)

which on my the attached screenshot from my laptop (which has approx 2GB RAM and has four Roxterms open) shows as:

1926904 - 1554004 = 372900 KiBytes

However, that value includes used buffers and cache, so a more indicative 'used' value (also provided by the MintPup free utility) is:

used - (buffers+cache) = 372900 - (54036 + 216700) = 102164 KiBytes

2. Mintpup 'top' utility. The MEM values displayed by this are pretty much identical to those of the 'free' utility.

3. Mintpup 'htop' utility. The MEM_used/MEM_total values displayed in the bar graph are in MiBytes, and the MEM_used figure is the same as free utility 'used - (buffers + cache) = 102164 KiBytes. However, MiBytes are KiBytes/1024, so my 'free' value 102164 shows as 102164/1024 = (approximately) 99 (the units being stated as MB, but that's a program error since should be stated as MiB). And MEM_total in htop shows as 1926904/1024 = (approximately) 1881 MiB.

NOTE: I've similarly checked these utilities in DebianDog Jessie 64 and also XenialDog32. The results and expected outputs are pretty much identical for DebianDog Jessie 64, but XenialDog is somewhat different...:

In XenialDog, the 'used' value (apparently accurately) reported by the 'free' utility is actually the /proc/meminfo values: MemTotal - MemFree - (Buffers + Cached), which should be the same as the MEM_used display given by htop (albeit divided by 1024 to convert it from KiBytes into MiBytes). However, unfortunately, htop in XenialDog seems to be all screwed up - htop display of MemTotal is also wrong...

Fred, from DebianDog thread tried MintPup's htop utility in XenialDog, which indicates lower memory usage, but, checking that against the XenialDog /proc/meminfo values (and its 'free' utility) indicates that MintPup htop doesn't give correct results in XenialDog either. Note that XenialDog 'top' utility report does match its 'free' utility report (i.e., unlike in MintPup, it subtracts buffers+cached from usage figure before reporting - but the 'top' utility report also seems reliable and accurate when compared to /proc/meminfo statistics).

Despite the bad news for XenialDog users about its htop utility (IMO) being all screwed up, the good news is that it is reporting far too high memory usage in my opinion. I believe, however, that the value given for RAM 'usage' in XenialDog 'free' utility is correct since it matches its /proc/meminfo reported values. That being so, XenialDog usage is actually LESS than MintPup immediately after booting! (XenialDog free reports 'usage' (which is MemTotal-MemFree-(Buffers+Cached)), immediately after boot on my 2GB RAM laptop, as approximately 86420 KiBytes = 86420/1024 MiBytes = 84 MiBytes (with the four Roxterms open and running the free, top and htop utilities in them), which is pretty good!!!
.
EDIT: XenialDog free buffers/cache and hence usage report not matching with my expectations from /proc/meminfo afterall:

http://www.murga-linux.com/puppy/viewto ... 887#903887

However, I think the figures work out for XenialDog free if you include /proc/meminfo stats for SReclaimable and SUnreclaim in buffers + cache. i.e. free utilities buff/cache = Buffers + Cached + SReclaimable + SUnreclaim

William

[saintless]

Thanks Pelo.

I will add something like that if I ever make new iso update:



I doubt anyone else is running MintPup on 64MB - 256MB RAM computers but I prefer not to take part in this RAM usage competition. What I find strange is reading here results using htop-mintpup in xenialdog, but not results running htop-xenial in mintpup. Seems logical to me to make such test before posting few pages with other RAM related data.

Poor screwed up Xenial htop - I welcome you with open arms in MintPup.

Toni

[starhawk]

saintless -- would you consider an XFCE version of MintPup...? I'd love to see a version of Mint that wasn't, you know, a gig plus in size, and I simply cannot stand JWM/ROX. (Color me weird, but I just can't make myself do it.)

[saintless]

...XFCE version of MintPup...?

There is OpenBox-XFCE setup from Fred. Link from the first post:

Click here for testing light OpenBox setup from Fred.

Just install and remaster.

Toni

[starhawk]

Openbox is LXDE

[saintless]

OK then. A little bit more work but anyone can do it:

mint ~ # apt-get install xfce4
Reading package lists... Done
Building dependency tree
Reading state information... Done
............................
The following NEW packages will be installed:
desktop-file-utils exo-utils gstreamer0.10-alsa gstreamer0.10-plugins-base
gtk2-engines-xfce iso-codes libexo-1-0 libexo-common libexo-helpers
libgarcon-1-0 libgarcon-common libgstreamer-plugins-base0.10-0
libgstreamer0.10-0 libgudev-1.0-0 libical1 libkeybinder0
libpolkit-gobject-1-0 libstartup-notification0 libthunarx-2-0
libunique-1.0-0 libupower-glib1 libwnck22 libxfce4ui-1-0 libxfce4ui-common
libxfce4ui-utils libxfce4util-common libxfce4util6 libxfconf-0-2
libxklavier16 libxres1 orage thunar thunar-data xfce4 xfce4-appfinder
xfce4-mixer xfce4-panel xfce4-session xfce4-settings xfconf xfdesktop4
xfdesktop4-data xfwm4
0 upgraded, 43 newly installed, 0 to remove and 7 not upgraded.
Need to get 10.1 MB/10.2 MB of archives.
After this operation, 58.3 MB of additional disk space will be used.
Do you want to continue? [Y/n] y

Then change /root/.xsession to execute xfce. Probabaly "exec xfce4-session"
And remaster. Should be easy for someone who knows XFCE. I have no experience with it.

Toni

[fredx181]

There is OpenBox-XFCE setup from Fred. Link from the first post:
saintless wrote:
Click here for testing light OpenBox setup from Fred.

Openbox, yes, but it's not XFCE, just openbox with tint2 etc..

Fred

[starhawk]

@ saintless -- I'll try it, but if it bombs I'll want help. I'm a /user/. Anything complex in a programming sort of way either works for me, or I need help with -- I'm a hardware guy. I can tinker with what I can touch. If it's software, it's almost inevitable that I'll need one of those yellow inflatable life rafts at some point early on in trying to whack something into functionality.

[mcewanw]

I prefer not to take part in this RAM usage competition. What I find strange is reading here results using htop-mintpup in xenialdog, but not results running htop-xenial in mintpup. Seems logical to me to make such test before posting few pages with other RAM related data.

Poor screwed up Xenial htop - I welcome you with open arms in MintPup.

Toni

Well its not a competition Toni. Not from my point of view anyway. I'm now a regular user of both MintPup (longtime user) and XenialDog, though recently mainly DDjessie64 because wanting to try out 64bit system. But I am also interested in Linux from technical point of view and it is painful when RAM usage tools are unpredictable in their reports and system efficiency has always been an attraction to the Puppy world. MintPup htop already yields the expected results, so no problem in MintPup whatsoever. Neither XenialDog free not its htop did, though I did track down the technical issue, which had been a concern of far more people than I interested from the technical point of view:

http://www.murga-linux.com/puppy/viewto ... 891#903891

As someone trying to help in system development I see it as one of my (less wanted) tasks to at least find out what is causing some of the usual tools of the job to not perform their function meaningfully. Of course I have better things to do too.

EDIT: and how does anyone know (without studying the source code) if a 'foreign' htop is reading the correct /proc/meminfo statistics for a different distribution from what it is intended. What it prints is meaningless without that knowledge, even if we like the values it prints because they suggest low RAM usage in the foreign distribution.

William

[saintless]

Hi William.

Well its not a competition Toni. Not from my point of view anyway.

It sounds like that to me:

I believe, however, that the value given for RAM 'usage' in XenialDog 'free' utility is correct since it matches its /proc/meminfo reported values. That being so, XenialDog usage is actually LESS than MintPup immediately after booting!

My main finding is that XenialDog actually uses less RAM on boot than MintPup (and backi's 'free' utility 'used' value is only 61252 KiBytes, which I think is the correct usage, which is quite low).

Do you really think 61252 KiBytes is the correct value in free command after starting Xorg with any WM?

EDIT: and how does anyone know (without studying the source code) if a 'foreign' htop is reading the correct /proc/meminfo statistics for a different distribution from what it is intended. What it prints is meaningless without that knowledge, even if we like the values it prints because they suggest low RAM usage in the foreign distribution.

When it shows more RAM usage in Xenial it is enough reason for assumptions like that:

EDIT: you are surely not suggesting that Linux Mint has patched htop to give more favourable looking RAM usage statistics?

When htop-mintpup shows less ram usage in xenialdog it behaves sane.
When htop-xenial shows less ram usage in mintpup it is screwed up.
Try to run free or top from Xenial in mintpup or check out ldd output for free and top in Xenial or Fedora23 and you will find who patched free command to give more favourable looking RAM usage.

Toni

[saintless]

I'll try it, but if it bombs I'll want help.

Post where it bombs then. I usually help if I can.
Just a hint if you use porteus-boot. The reboot and shutdown commands in XFCE menu must be changed to execute wmreboot and wmpoweroff instead if you like to use save on Exit boot code. I don't expect other bombs.

Toni

[mcewanw]

Hi William.

Well its not a competition Toni. Not from my point of view anyway.

It sounds like that to me:
Toni

Forget it, that is just nonsense talk. Why on earth would I favour one of these distributions over another. I don't - end of story.

William

[fredx181]

Hi Toni, William,

mcewanw wrote:
Well its not a competition Toni. Not from my point of view anyway.

It sounds like that to me:

Well, I do not have intention of competition at all and I'm sure William doesn't. It's about gathering knowledge and fair comparison of RAM usage .
Started for me when I got reports of very high RAM usage in Xenial reported from htop, it's a pity that the discussion took place (too much) in this thread.
But, Toni, I can imagine you got a bitter taste of competition (reading the last few pages here) from some of Pelo's comments.

Fred

[starhawk]

I'll try it, but if it bombs I'll want help.

Post where it bombs then. I usually help if I can.
Just a hint if you use porteus-boot. The reboot and shutdown commands in XFCE menu must be changed to execute wmreboot and wmpoweroff instead if you like to use save on Exit boot code. I don't expect other bombs.

Toni

Thanks! I'll keep you posted... may or may not be today, I got errands to run today. But I'll keep you posted.

[mcewanw]

Hi Toni and Fred,

I think it would be a good idea to resurrect a thread (only one would do) for DD/MintPup/Ubuntu development work where we are free to make any comments or post we like regarding development work we have spent time doing. There should be no place for accusations, one way or the other - we all, to some extent, have different interests, views, and motivations and no one has a right to demotivate anyone else.

There may well at times be pages of development efforts, and often with errors or faulty statements in them - if they were immediately perfect, that would be great, but development is all about knowledge seeking and improvement - which requires publishing work with the open idea that others can contribute and discover and repair errors.

To keep users and other developers notified of development posts in that separate thread it would be enough to simply make a simple post in the relevant users' thread stating what the development post is about and a link to it.

Of course, if development contributions are no longer required or considered irrelevant or annoying in any way, I am happy to leave these threads to yourselves since I am only now an occasional contributor anyway. To a large extent everything is running quite smoothly now so good just to enjoy the close to polished products.

William

[saintless]

Hi Fred.

It's about gathering knowledge and fair comparison of RAM usage .

Posting screenshot here from mint-htop in Xenial is not fair. You should also post the result from mint-free and mint-top commands in Xenial. And the result from same Xenial commands in Mint. And you should try to find information how is /proc/meminfo generated in Xenial and Mint.
If you do that you should ask yourself what could be the possible reason free and top commands in Xenial to have systemd as dependency. I can't imagine why systemd has to be installed in order to use free and top commands, can you?
If you think htop from Xenial is screwed up (I don't think so) then fair is to create bug report to the developers. Maybe they have the answer to your question. But instead you are writing your own script for fair comparison.
So the question is fair to whom, Fred? Do you think it is fair to Linux Mint what google search gives for this:

Code: Select all

linux mint htop patched

But, Toni, I can imagine you got a bitter taste of competition (reading the last few pages here) from some of Pelo's comments.

Please, don't put the blame on Pelo. He was not the one who posted here screenshot from mint-htop running in XenialDog.
I don't know what this screenshot has to do with MintPup but it ended up with MintPup - XenialDog RAM usage comparison in this thread. Something I wouldn't even think to do since they are based on different ubuntu versions (Trusty and Xenial), different init (upstart and systemd), different kernel, different WM, different start up scripts and applications.
In old computers up to 256RAM you don't need htop to tell you how much RAM is in use. You can feel it by how responsive the system is in each moment. Who cares what htop shows if you can't feel the system slower having 4Gb RAM or more? And who cares if htop says 50Mb RAM usage in your screenshot if I can't boot your distro on my old 256Mb ram machine?

Toni

[saintless]

Hi William.

I think it would be a good idea to resurrect a thread (only one would do) for DD/MintPup/Ubuntu development work where we are free to make any comments or post we like regarding development work we have spent time doing.

There is such thread we used for all DebianDog versions development. No problem in my opinion to mix it with MintPup, XenialDog:
http://murga-linux.com/puppy/viewtopic. ... start=4230
My last answer is a year ago but this is only because I see no questions or new posts there. If you start posting something there the answers will come there too I guess.
Also github issues pages are good place.

Toni

[fredx181]

Please, don't put the blame on Pelo. He was not the one who posted here screenshot from mint-htop running in XenialDog.

Blame!? am I to blame turning things in a competition?
A competition in my eyes is when two or more opponents are trying to win.
So.. Am I trying to win something? No.

I think you are misunderstanding or maybe I misunderstood when you used the word "competition".

Posting the screenshot was just a beginning of a search for fair comparison of RAM usage between distros.
Quite naive, yes, you are right about all the other factors required for a fair comparison.

Started with report from backi in Xenialdog thread which worried me about how to trust htop on Xenialdog:

Htop shows 362 megs of ram usage in Htop
( really.... 362 in Htop )

Which I find strange (BTW, I never said that htop in Xenial is screwed)
For me htop output is much less than that in xenialdog, but there's a big difference between free and htop.
Still don't know what's right, free shows too little IMO. (61MB)

As I said, I regret that this discussion took place in this thread, so OK for me to continue in the Light-Wheezy thread.
But I have no interest if you still think I'm looking for some competition.

Edit: Also I'd like to mention (in case misunderstanding) that I don't think the htop from Mintpup has a more favorable output (I didn't write about patched or something, BTW).
I just think it's realistic (and very similar to free, lxtask, conky).

EDIT:

If you think htop from Xenial is screwed up (I don't think so) then fair is to create bug report to the developers. Maybe they have the answer to your question. But instead you are writing your own script for fair comparison.

My purpose of these scripts are; they are exactly doing/simulating what is the (different) 'free' output on Xenialdog and Debiandog for investigating and to share. (and I just wrote my thoughts about it)
And, as I said earlier in PM, I don't like your aggressive tone in this, Toni, it's misplaced, I get the feeling from it we are having some battle here, which should not be like that.

Fred

[mcewanw]

Hi William.
There is such thread we used for all DebianDog versions development. No problem in my opinion to mix it with MintPup, XenialDog:
http://murga-linux.com/puppy/viewtopic. ... start=4230
My last answer is a year ago but this is only because I see no questions or new posts there. If you start posting something there the answers will come there too I guess.
Also github issues pages are good place.

Toni

Hi Toni,

Unfortunately, though I first suggested using github, I haven't myself been keeping up with that at all, though I shall try to get back up to speed on that again. In the meantime, should I wish to post anything DebianDog/Mintpup/Ubuntu related to development I will use that old thread you mention.

I honestly think you are on the wrong track thinking Fred (or I) have posted anything with some competition in mind. There is a noted issue in the new XenialDog in terms of htop reporting unexpectedly high RAM usage, and that of course could put people off trying it, which I therefore felt was worth trying to get to the bottom of.

I consider all these distributions created by Fred and yourself from the same 'house' and personally feel support for all of them equally. I still inherently prefer JWM to Openbox, which is why I pushed Fred recently to create a JWM menu version for his Jessie64. Also I like having xdm working even though I am also generally auto-logging-in as root. We all have our own favourites, I am sure - mine really being MintPup, albeit remastered to the way I like it.

I think you should bite your tongue and realise that you are imagining competition when there is none. But that is up to you!

William

EDIT: By the way Fred, I was attempting some humour when I asked half-heartedly if you thought Linux Mint htop was patched to give good report. Poor attempt no doubt. But actually that conky github link I gave clearly states that some of the Opensuse versions of these utilities had been patched. That was to alter what /proc/meminfo stats are counted. Not sure what the inference or big deal is about all this myself.

EDIT2: And it was just my opinion, and remains so, that htop in Xenial is 'screwed' in the sense that it reports far too high RAM usage compared to previous htop reports we are used to. Perhaps Ubuntu publish the new meaning somewhere, but until and unless people are made aware of the change, then its report has no meaning and thus screwed as a useful tool until then - for me at least, unless some arbitrary result means something to you???

http://idioms.thefreedictionary.com/screwed+up