App::SpeedTest(3)
π App::SpeedTest π
okay so bestie, quick vibe check: this is a manpage from the User Contributed Perl Documentation section, and its whole name-title situation is App::SpeedTest β that's what we're reading today, and honestly? it's giving "useful little tool that absolutely ate." π
Let's get into it.
πΈ NAME πΈ
App::SpeedTest β a command-line interface to speedtest.net. yes, the actual speedtest dot net. from your terminal. no flash, no browser, no nonsense. β¨
β¨ SYNOPSIS β¨
$ speedtest [ --no-geo | --country=NL ] [ --list | --ping ] [ options ]
$ speedtest --list
$ speedtest --ping --country=BE
$ speedtest
$ speedtest -s 4358
$ speedtest --url=http://ookla.extraip.net
$ speedtest -q --no-download
$ speedtest -Q --no-upload
like, that's the whole energy right there β pick your server sitch, and go πββοΈπ¨
π DESCRIPTION π
okay so the provided perl script is a command-line interface to the speedtest.net <http://www.speedtest.net/> infrastructure, and the whole point is that flash is not required ππ¦
basically it was written to feature the functionality that speedtest.net offers β but without the overhead of flash or java, and without needing a browser at all. iconic behavior, no cap. π
π¦ Raison-d'Γͺtre π¦
the whole reason this tool exists: it's here to give you a quick indication of the achievable throughput of your current network. and ngl, that throughput can drop dramatically if you're behind (several) firewalls, or badly configured networks β or network parts like switches, hubs and routers. π like, the vibes can be immaculate on paper and then your actual speed is in shambles.
it was inspired by speedtest-cli <https://github.com/sivel/speedtest-cli>, a project written in python. but β and I say this with love β I neither like python, nor did I like the default behavior of that script. π I also think it does not take the right decisions in choosing the server based on distance instead of speed. and that actually matters if one has fiber lines, bestie. I prefer speed over distance, periodt. π₯
π Command-line Arguments π
okay, here's the whole menu. every single one, no skipping π
-?,--helpβ show all available options, then exit. the "help me help you" button π«Ά-V,--versionβ show program version and exit. identity check π--manβ show the builtin manual usingpod2manandnroff. the fancy formatted one πΈ--infoβ show the builtin manual usingpod2text. the plain-text one π-v[#],--verbose[=#]β set verbose level. the default value is1, and a plain-vwithout a value will set the level to2. louder = more tea β--simpleβ an alias for-v0. quiet queen energy π€«--allβ no (default) filtering on available servers. super useful when finding servers outside of the country of your own location. the passport-free option π-g,--geoβ use GEO-IP service to find the country your ISP is located. the default is true. if you disable it (--no-geo), the server to use will be based on distance instead of on latency. choose your fighter βοΈ-cXX,--cc=XX,--country=XXβ pass the ISO country code to select the servers πΊοΈ
$ speedtest -c NL ...
$ speedtest --cc=B ...
$ speedtest --country=D ...
--list-ccβ fetch the server list and then show the list of countries the servers are located with their country code and server count. it's literally a world tour in list form βοΈ
$ speedtest --list-cc
AD Andorra 1
AE United Arab Emirates 4
:
ZW Zimbabwe 6
you can then use that code to list the servers in the chosen country, as described below. iykyk π
-l,--listβ this option will show all servers in the selection with the distance in kilometers to the server. cute little table, very demure π
$ speedtest --list --country=IS
1: 10661 - Tengir hf Akureyri 1980.02 km
2: 21605 - Premis ehf ReykjavΓk 2039.16 km
3: 3684 - Nova Reykjavik 2039.16 km
4: 6471 - Gagnaveita Reykjavikur Reykjavik 2039.16 km
5: 10650 - Nova VIP Reykjavik 2039.16 km
6: 16148 - Hringidan Reykjavik 2039.16 km
7: 4818 - Siminn Reykjavik 2039.16 km
8: 17455 - Hringdu ReykjavΓk 2039.16 km
9: 4141 - Vodafone ReykjavΓk 2039.16 km
10: 3644 - Snerpa Isafjordur 2192.27 km
-p,--ping,--ping=40β show a list of servers in the selection with their latency in ms. β±οΈ be very patient if running this with--allβ like, go make a snack, bestie π
$ speedtest --ping --cc=BE
1: 4320 - EDPnet Sint-Niklaas 148.06 km 52 ms
2: 12627 - Proximus Brussels 173.04 km 55 ms
3: 10986 - Proximus Schaarbeek 170.54 km 55 ms
4: 15212 - Telenet BVBA/SPRL Mechelen 133.89 km 57 ms
5: 29238 - Arcadiz DIEGEM 166.33 km 58 ms
6: 5151 - Combell Brussels 173.04 km 59 ms
7: 26887 - Arxus NV Brussels 173.04 km 64 ms
8: 4812 - Universite Catholiq⦠Louvain-La-Neuv 186.87 km 70 ms
9: 2848 - Cu.be Solutions Diegem 166.33 km 75 ms
10: 12306 - VOO Liège 186.26 km 80 ms
11: 24261 - Une Nouvelle Ville⦠Charleroi 217.48 km 147 ms
12: 30594 - Orange Belgium Evere 169.29 km 150 ms
if a server does not respond, a very high latency is used as default. rude of the server, honestly π
this option only shows the 40 nearest servers. the number can be changed as optional argument ποΈ
$ speedtest --cc=BE --ping=4
1: 4320 - EDPnet Sint-Niklaas 148.06 km 53 ms
2: 29238 - Arcadiz DIEGEM 166.33 km 57 ms
3: 15212 - Telenet BVBA/SPRL Mechelen 133.89 km 62 ms
4: 2848 - Cu.be Solutions Diegem 166.33 km 76 ms
-1,--one-lineβ generate a very short report easy to paste in e.g. IRC channels. tiny, punchy, perfect for the group chat π¬
$ speedtest -1Qv0
DL: 40.721 Mbit/s, UL: 30.307 Mbit/s
-B,--bytesβ report throughput in Mbyte/s instead of Mbit/s. bytes girlies rise π§-C,--csvβ generate the measurements in CSV format. the data can be collected in a file (by a cron job) to be able to follow internet speed over time. spreadsheet slay πβ¨
the reported fields are:
- A timestam (the time the tests are finished)
- The server ID
- The latency in ms
- The number of tests executed in this measurement
- The direction of the test (D = Down, U = Up)
- The measure avarage speed in Mbit/s
- The minimum speed measured in one of the test in Mbit/s
- The maximum speed measured in one of the test in Mbit/s
$ speedtest -Cs4358
"2015-01-24 17:15:09",4358,63.97,40,D,93.45,30.39,136.93
"2015-01-24 17:15:14",4358,63.97,40,U,92.67,31.10,143.06
-U,--skip-undefβ skip reporting measurements that have no speed recordings at all. the default is to report these as0.00 .. 999999999.999. yeah, that's the whole range, don't ask π-P,--prtgβ generate the measurements in XML suited for PRTG π§Ύ
$ speedtest -P
<?xml version="1.0" encoding="UTF-8" ?>
<prtg>
<text>Testing from My ISP (10.20.30.40)</text>
<result>
<channel>Ping</channel>
<customUnit>ms</customUnit>
<float>1</float>
<value>56.40</value>
</result>
<result>
<channel>Download</channel>
<customUnit>Mbit/s</customUnit>
<float>1</float>
<value>38.34</value>
</result>
<result>
<channel>Upload</channel>
<customUnit>Mbit/s</customUnit>
<float>1</float>
<value>35.89</value>
</result>
</prtg>
--url[=XXX]β with no value, show server url in list. π with a value, use that specific server url: it will not scan available servers. main character server selection π―--ipβ show IP for server. who are you really π-T[#],--try[=#]β use the top # (based on lowest latency or shortest distance) from the list to do all required tests. the "give me options" mode π
$ speedtest -T3 -c NL -Q2
Testing for 80.x.y.z : XS4ALL Internet BV (NL)
Using 13218: 26.52 km 25 ms XS4ALL Internet BV
Test download .. Download 31.807 Mbit/s
Test upload .. Upload 86.587 Mbit/s
Using 15850: 26.09 km 25 ms QTS Data Centers
Test download .. Download 80.763 Mbit/s
Test upload .. Upload 77.122 Mbit/s
Using 11365: 26.09 km 27 ms Vancis
Test download .. Download 106.022 Mbit/s
Test upload .. Upload 82.891 Mbit/s
Rank 01: Server: 11365 26.09 km 27 ms, DL: 106.022 UL: 82.891
Rank 02: Server: 15850 26.09 km 25 ms, DL: 80.763 UL: 77.122
Rank 03: Server: 13218 26.52 km 25 ms, DL: 31.807 UL: 86.587
$ speedtest -1v0 -T5
DL: 200.014 Mbit/s, UL: 159.347 Mbit/s, SRV: 13218
DL: 203.599 Mbit/s, UL: 166.247 Mbit/s, SRV: 15850
DL: 207.249 Mbit/s, UL: 134.957 Mbit/s, SRV: 11365
DL: 195.490 Mbit/s, UL: 172.109 Mbit/s, SRV: 5972
DL: 179.413 Mbit/s, UL: 160.309 Mbit/s, SRV: 2042
Rank 01: Server: 15850 26.09 km 30 ms, DL: 203.599 UL: 166.247
Rank 02: Server: 5972 26.09 km 32 ms, DL: 195.490 UL: 172.109
Rank 03: Server: 13218 26.52 km 23 ms, DL: 200.014 UL: 159.347
Rank 04: Server: 11365 26.09 km 31 ms, DL: 207.249 UL: 134.957
Rank 05: Server: 2042 51.41 km 33 ms, DL: 179.413 UL: 160.309
-s#,--server=#,--server=filenameβ specify the ID of the server to test against. this ID can be taken from the output of--listor--ping. using this option prevents fetching the complete server list and calculation of distances. it also enables you to always test against the same server. loyalty, but for servers π«Ά
$ speedtest -1s4358
Testing for 80.x.y.z : XS4ALL Internet BV ()
Using 4358: 52.33 km 64 ms KPN
Test download ........................................Download: 92.633 Mbit/s
Test upload ........................................Upload: 92.552 Mbit/s
DL: 92.633 Mbit/s, UL: 92.552 Mbit/s
this argument may be repeated to test against multile servers, more or less like specifying your own top x (as with -T). the more the merrier β¨
$ speedtest -s 22400 -s 1208 -s 13218
Testing for 185.x.y.z : Freedom Internet BV ()
Using 13218: 80.15 km 32 ms XS4ALL Internet BV
Test download ........................................Download 66.833 Mbit/s
Test upload ........................................Upload 173.317 Mbit/s
Using 1208: 51.19 km 37 ms Qweb | Full-Service Hosting
Test download ........................................Download 52.077 Mbit/s
Test upload ........................................Upload 195.833 Mbit/s
Using 22400: 80.15 km 46 ms Usenet.Farm
Test download ........................................Download 96.341 Mbit/s
Test upload ........................................Upload 203.306 Mbit/s
Rank 01: Server: 22400 80.15 km 46 ms, DL: 96.341 UL: 203.306
Rank 02: Server: 1208 51.19 km 37 ms, DL: 52.077 UL: 195.833
Rank 03: Server: 13218 80.15 km 32 ms, DL: 66.833 UL: 173.317
Okay so real talk about that filename β if you pass one, it's gotta look like the server-like structure you'd get from the speedtest server-list, maybe filled out with upload- and download URL's. And like: you can only pass one filename that isn't all digits. If you do pass one, every remaining "-s" argument just gets ignored. Sorry bestie, that's the rules π
{ cc => "NL",
country => "Netherlands",
host => "unlisted.host.amsterdam:8080",
id => 9999,
lat => "52.37316",
lon => "4.89122",
name => "Amsterdam",
ping => 20.0,
sponsor => "Dam tot Damloop",
url => "http://unlisted.host.amsterdam/speedtest/speedtest/upload.php",
url2 => "http://unlisted.host.amsterdam/speedtest/speedtest/upload.php",
dl_list => [
"http://unlisted.host.amsterdam/files/128.bin",
"http://unlisted.host.amsterdam/files/256.bin",
# 40 URL's pointing to files in increasing size
"http://unlisted.host.amsterdam/files/2G.bin",
],
ul_list => [
# 40 URL's
],
}
-t#,--timeout=#β sets the maximum timeout, in seconds. Tick tock β³-d,--downloadβ runs the download tests. This is the default unless--uploadgets passed. β¨-u,--uploadβ runs the upload tests. This is the default unless--downloadgets passed. π«Ά-q[#],--quick[=#],--fast[=#]β don't run the full test, bestie. The default test runs 40 tests, sorted on increasing test size (and thus test duration), and those longer tests can take forever on slow connections without adding any value π. The default value for-qis 20, but any value between 1 and 40 is allowed.-Q[#],--realquick[=#]β same deal, don't run the full test. Again: the default test runs 40 tests, sorting on increasing test size (and thus test duration), and long(er) tests may take too long on slow connections without adding value. The default value for-Qis 10, but any value between 1 and 40 is allowed. β‘-mXX,--mini=XXβ runs the speedtest on a speedtest mini server. π§--source=XXβ NYI, literally just mentioned for speedtest-cli compatibility. Don't come for me π
π EXAMPLES π
See "SYNOPSIS" and Command-line arguments. That's the whole vibe, go look π
π¦ DIAGNOSTICS π¦
...
(Yeah. That's it. Iconic of them honestly.)
π BUGS and CAVEATS π
Because of how the language is implemented, it may report speeds that are not consistent with the speeds reported by the web interface or other speed-test tools. No cap. Same energy for the reported latencies β those are not to be compared to the ones tools like ping report, so don't @ me π
π TODO π
Improve documentation
What did I miss? π
Enable alternative XML parsers
XML::Simple is not the recommended XML parser, but it sufficed on startup. All the other API's are more complex. It's giving "it works, leave it alone" π«
β PORTABILITY β
Perl has been ported to a whole plethora of operating systems, so this CLI will work fine on all systems that fulfill the requirement as listed in Makefile.PL (or the various META files). Slay, cross-platform queen π
The script has been tested on Linux, HP-UX, AIX, and Windows 7. Aged like fine wine, fr fr.
Debian wheezy will run with just two additional packages:
# apt-get install libxml-simple-perl libdata-peek-perl
π₯ SEE ALSO π₯
As an alternative to speedtest.net <http://www.speedtest.net/>, you could consider <http://compari.tech/speed>.
And the speedtest-cli <https://github.com/sivel/speedtest-cli> project β the one that inspired me to improve a broken CLI written in python into our beloved language Perl. Character development at its finest π
πΈ CONTRIBUTING πΈ
General
I am always open to improvements and suggestions β use issues at github issues <https://github.com/Tux/speedtest/issues>. Come through! π«Ά
Style
Heads up: I will never accept a pull request that does not strictly conform to my style, however much you might hate it. You can read the reasoning behind my preferences here <https://tux.nl/style.html>.
I really don't care about mixed spaces and tabs in (leading) whitespace. Lowkey generous of me ngl π
π WARRANTY π
This tool is by no means a guarantee to show the correct speeds. It's only to be used as an indication of the throughput of your internet connection, and the values shown cannot be used in a legal debate. So don't bring these numbers to court, bestie β they will not hold up π
π AUTHOR π
H.Merijn Brand <linux@tux.freedom.nl> wrote this for his own personal use, but was asked to make it publicly available as application. Main character energy, honestly β¨
π COPYRIGHT AND LICENSE π
Copyright (C) 2014-2026 H.Merijn Brand
This software is free; you can redistribute it and/or modify it under the same terms as Perl itself.
And that's the whole page, bestie π This beauty shipped as perl v5.42.2 on 2026-03-15 β App::SpeedTest(3), go forth and speed-test responsibly β¨π
π¦
π¦ where this page comes from
- alpine 3.24: perl-app-speedtest-doc 0.31-r0 Β· GPL-1.0-or-later OR Artistic-1.0-Perl
- π girlified by deepseek-ai/DeepSeek-V4.1-Flash-1017 on : a language model's rewrite of the original, under its terms