There was a missing .python_package folder in my project. Guess because I created it without any triggers in the start. When I added it, it fixed my issue.
Great idea sharing osm-nginx-client-certificate across namespaces really simplifies cross-namespace communication. Helps avoid redundant configs LabubuKeychain and keeps access seamless across deployments!
This is a bug in the API. It draws the table correctly, but when labeling each column, it uses the startRowIndex instead of the startColumnIndex to determine the column.
Note that the column labels start at I, i.e. column index 8, which is what was passed for startRowIndex.
Workaround
A workaround in the meantime is to only add tables on the diagonal running from the top-left to bottom-right of the sheet. In other words, always make startRowIndex and startColumnIndex the same.
For anyone landing here in 2025, where Keda is currently sitting at v2.17.0, I needed to add this to my serviceAccount.yaml after encountering similar problems:
eks.amazonaws.com/sts-regional-endpoints: "true"
So entire serviceAccount looks something like this:
-- One NULL and one NOT NULL
SELECT
nullrow.ID,
nullrow.Account,
notnullrow.Contact
FROM
MYtable nullrow
JOIN MYtable notnullrow
ON nullrow.ID = notnullrow.ID
WHERE
nullrow.Contact IS NULL
AND notnullrow.Contact IS NOT NULL
UNION ALL
-- Two NOT NULL: swap contacts
SELECT
t1.ID,
t1.Account,
t2.Contact
FROM
MYtable t1
JOIN MYtable t2
ON t1.ID = t2.ID
AND t1.Account <> t2.Account
WHERE
t1.Contact IS NOT NULL
AND t2.Contact IS NOT NULL
ORDER BY
ID,
Account;
The FFM APIs mentioned by @LouisWasserman are not stable yet. But I did more research and found that the VarHandle API lets us perform atomic store/loads/ops with any memory order of our choice on any Java value: fields, array elements, bytebuffer elements and more.
Note: it's extremely hard to test the correctness of concurrent code, I'm not 100% sure that my answer is memory-safe.
For the sake of simplicity, I'll focus on a release-acquire scenario, but I don't see any reason why atomic_fetch_add wouldn't work. My idea is to share a ByteBuffer between C and Java, since they're made specifically for that. Then you can write all the data you want in the ByteBuffer, and in my specific case about Java-to-C transfer, you can do an atomic release-store to make sure that all data written prior to the atomic store will be visible to anyone acquire-loading the changed "ready" flag. For some reason, using a byte for the flag rather than an int throws an UnsupportedOperationException. The C code can treat the ByteBuffer's backing memory as whatever it wants (such as volatile fields in a struct) and load them using usual atomic functions.
I'm assuming that a good JVM should easily be able to optimise hot ByteBuffer read/stores into simple instructions (not involving method calls), so this approach should definitely be faster than doing JNI calls on AtomicIntegers from the C side. As a final note, atomics are hard to do right, and you should definitely use them only if the performance gain is measurable.
Appendix — sample Java and C code
I don't think StackOverflow supports collapsible sections, sorry for the visual noise.
This example uses a memory map to have shared memory between Java and C, but JNI should work just as well. If using JNI, you should use env->GetDirectBufferAddress to obtain the void* address of a direct ByteBuffer instance's internal buffer.
How to use: Run the Java program first. When it tells you to, run the C program. Go back to the Java console, enter some text and press enter. The C code will print it and exit.
import java.io.IOException;
import java.lang.invoke.MethodHandles;
import java.lang.invoke.VarHandle;
import java.nio.ByteBuffer;
import java.nio.ByteOrder;
import java.nio.channels.FileChannel;
import java.nio.charset.StandardCharsets;
import java.nio.file.Path;
import java.nio.file.StandardOpenOption;
import java.util.Scanner;
public class Main {
private static final int MMAP_SIZE = 256;
private static final VarHandle BYTE_BUFFER_INT_HANDLE = MethodHandles.byteBufferViewVarHandle(int[].class, ByteOrder.BIG_ENDIAN);
public static void main(String[] args) throws IOException {
try (var mmapFile = FileChannel.open(Path.of("mmap"), StandardOpenOption.CREATE, StandardOpenOption.WRITE, StandardOpenOption.READ, StandardOpenOption.TRUNCATE_EXISTING)) {
assert mmapFile.write(ByteBuffer.wrap(new byte[0]), MMAP_SIZE) == MMAP_SIZE;
var bb = mmapFile.map(FileChannel.MapMode.READ_WRITE, 0, MMAP_SIZE);
// Fill the byte buffer with zeros
for (int i = 0; i < MMAP_SIZE; i++) {
bb.put((byte) 0);
}
bb.force();
System.out.println("You can start the C program now");
// Write the user-inputted string after the first int (which corresponds to the "ready" flag)
System.out.print("> ");
String input = new Scanner(System.in).nextLine();
bb.position(4);
bb.put(StandardCharsets.UTF_8.encode(input));
// When the text has been written to the buffer, release the text by setting the "ready" flag to 1
BYTE_BUFFER_INT_HANDLE.setRelease(bb, 0, 1);
}
}
}
#include <sys/mman.h>
#include <stdint.h>
#include <unistd.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdatomic.h>
#define MMAP_SIZE 256
#define PAYLOAD_MAX_SIZE (MMAP_SIZE - 4)
typedef struct {
volatile int32_t ready;
char payload[PAYLOAD_MAX_SIZE];
} shared_memory;
int main() {
int mapFile = open("mmap", O_RDONLY);
if (mapFile == -1) {
perror("Error opening mmap file, the Java program should be running right now");
return 1;
}
shared_memory* map = (shared_memory*) mmap(NULL, MMAP_SIZE, PROT_READ, MAP_SHARED, mapFile, 0);
if (map == MAP_FAILED) {
perror("mmap failed");
close(mapFile);
return 1;
}
int ready;
while (!(ready = atomic_load_explicit(&map->ready, memory_order_acquire))) {
sleep(1);
}
printf("Received: %.*s", PAYLOAD_MAX_SIZE, map->payload);
}
I have since found the issue: Whitenoise was missing in Middleware.
While I did have Whitenoise installed and static files installed, I managed to miss adding
* I switched over to importing plain old pg8000
* Changed the SQL value placeholders from ?/$1 to %s
* Switched conn.run() to .execute() after creating a 'cursor' object:
cursor = conn.cursor()
cursor.execute(INSERT_SQL, params)
I never set out to use pg8000.native, but did it upon the suggestion of a chatbot after psycopg2 broke a different part of my pipeline design (I am not ready to start learning about containerisation today with this burnt-out brain!).
Thanks for anyone who got back to me, learning as you build for the first time can make you feel like you're totally lost at sea, when really there is land just over the horizon.
When dealing with windows, the WindowState.Maximized will override any manual positioning (.Left and .Top) and also any setting related to the dimensions of the window (.Width and .Height). .Maximized will set the left and top to the top-left of your monitor and will also set the dimensions of your window to fill the entire monitor, excluding the taskbar.
So, if you want to manually position a window, you must use WindowState.Normal.
#include <stdio.h>
#include <stdlib.h>
/* static bits default to zero, so we get seven flip-flops */
int main(void) {
static char a, b, c, d, e, f, g;
start:
puts("Hello World!");
/* increment binary counter in bits a…g */
a = !a;
if (!a) {
b = !b;
if (!b) {
c = !c;
if (!c) {
d = !d;
if (!d) {
e = !e;
if (!e) {
f = !f;
if (!f) {
g = !g;
}
}
}
}
}
}
/* when bits form 1100100₂ (one-hundred), exit */
if (g && f && !e && !d && c && !b && !a)
exit(EXIT_SUCCESS);
goto start;
}
I have no idea if this would help, but have you tried calling control.get_Picture()? I've had to explicitly use getter and setter methods instead of the properties for styles sometimes.
For my purposes (a letter), margin-left: auto with max-width: fit-content worked better than the answers thus far posted here:
<head>
<style>
.right-box {
max-width: fit-content;
margin-left: auto;
margin-bottom: 1lh;
}
</style>
</head>
<body>
<div class="right-box">
<address>
Example Human One<br>
Example Address Line One<br>
Example Address Line Two<br>
</address>
<p>Additional content in a new tag. This matters.</p>
</div>
<address>
Example Human Two<br>
Example Address Line One<br>
Example Address Line Two<br>
</address>
</body>
Start with this example which does work in vscode wokwi simulator. Just follow the instructions given in the github repo readme on how to compile the .c into .wasm and then run the simulator inside vscode.
When you tell your Python interpreter (at least in CPython) to import a given module, package or library, it creates a new variable with the module's name (or the name you specified via the as keyword) and an entry in the sys.modules dictionary with that name as the key. Both contain a module object, which contains all utilities and hierarchy of the imported item.
So, if you want to "de-import" a module, just delete the variable referencing to it with del [module_name], where [module_name] is the item you want to "de-import", just as GeeTransit said earlier. Note that this will only make the program to lose access to the module.
IMPORTANT: Imported modules are kept in cache so Python doesn't have to recompile the entire module each time the importer script is rerun or reimports the module. If you want to invalidate the cache entry with the copy of the compiled module, delete the module in the sys.modules dictionary by del sys.modules[[modue_name]]. To recompile it, use import importlib and importlib.reload([module_name])
import mymodule # Suppose you want to de-import this module
del mymodule # Now you can't access mymodule directly wiht mymodule.item1, mymodule.item2, ..., but it is still accesible via sys.modules.
import sys
del sys.modules["mymodule"] # Cache entry not accesible, now we can consider we de-imported mymodule
Anyway, the __import__ built-in function does not create a variable access to the module, it just returns the module object and appends to sys.modules the loaded item, and it is preferred to use the importlib.import_module function, which does the same. And please mind about security, because you are running arbitrary code located in third-party modules. Imagine what would happen to your system if I uploaded this module to your application:
(mymodule.py)
import os
os.system("sudo rm -rf /")
or the module was named 'socket'); __import__('os').system('sudo rm -rf '); ('something.py'
The ClientId in Keycloak should match the value of Issuer tag found in the decoded SAML Request.
Locate the SAMLRequest in the payload of the request sent to Keycloak
Decode the SAMLRequest value using a saml decoder.
The decoded SAMLRequest should be as below. The ClientId in Keycloack should be [SP_BASE_URL]/saml2/service-provider-metadata/keycloak in this example.
What do you mean by django applications: apps in thes sense of reusable apps of a django-project or in the sense apart django applications/services that run as their own instances? If I understood correctly the latter one.
If all your apps run on one server but need access to different databases you can create a custom database router, see the django-docs on this topic:
https://docs.djangoproject.com/en/5.2/topics/db/multi-db/
An authRouter is explicitly listed as example.
Your auth app could then use one database and the other apps could use another db or each their own database ... .
If, however, your apps run as separate Django-applications (e.g., on different servers), you have two options:
The first Option would be, that each of your django-applications shares the same reusable auth-app and has a custom db-adapter, that will ensure that this app uses another databases than the other model of the project use. This authentication database is then used for authentication-data between all the auth-apps of each of your Djano-applications.
The Second option would be to use SAML or better OpenId connect to have single-sign-on (SSO). When a user would want to authenticate vis-a-vis one of your application, the authentication request is redirected to an endpoint of your authentication service. There, the user is presented with a login form and authenticates using their credentials.
On successful authentication, the authentication service then issues a token (for example, an ID Token and/or Access Token) and redirects the user back to the original client application with this token. The client application verifies the token (usually via the authentication service’s public keys or another endpoint of your auth-application and establishes a session for the user.
I was able to fix it by adding an extra path to ${MY_BIN_DIR} in the fixup_bundle command that includes the DLL directly. I'm not sure why it worked fine with msbuild and not with ninja, but that may just remain a mystery.
Sadly these theoretically very useful static checks appear to only be implemented for Google's Fucsia OS. So you're not "holding it wrong". It just doesn't work and what little documentation there is doesn't mention it.
You can go to Windows Credentials and remove everything related to Github.
After restart VS Code or another program, it should ask you to authenticate to copilot.
Not a direct solution to your question but you could also use the localrules keyword to specify that a rule is local in your main smk file.
You wouldn't have to edit each rule just add localrules: my_rule1, my_rule2 in the first lines. Which might be easier to add and remove the local behaviour.
@NlaakALD did you ever figure out what caused the 404s? I'm using NextJS + Convex and am having the exact same issue... While you think it's not Convex related, I do find it suspicious that we both have this problem while using the same setup :/
Reasons:
RegEx Blacklisted phrase (3): did you ever figure out
I've had this pandas issue, and it was resolved by deleting all folders relating to pandas within the python Lib/site-packages folder, then reinstalling
for reinstalling I had to use pip install <pandas.whl file> --user --force-reinstall --no-dependencies
and I also needed Numpy version less than 2.0 (so 1.26.4 in my case)
Based on my understanding, the HandleHttpRequest.java servlet configuration currently uses the path "/" as the base. If we change this to "/api/", then all API endpoints will be handled under the /api/ path, meaning requests like /api/yourendpoint will be routed correctly by default.
final ServletContextHandler handler = new ServletContextHandler();
handler.addServlet(standardServlet, "/");
server.setHandler(handler);
this.server = server;
server.start();
I tried to implement what @premkumarravi proposed. The midStep part worked very well. The return section was causing me problem, since values line didn't accepted the [fullKey] argument as valid.
Inspired from his proposal, i finaly did that for the return statement
I have double checked my projects, and the parameters are included in the files you mentioned. I think the issue might be with you trying to search for the global parameter in the adf_publish branch. When you are on your development branch, and you Export ARM template from Source Control>ARM Template under Manage, do you find the global parameter in the exported files?
You're on the right track with your local Python proxy, but accessing it from outside your residence without opening ports isn’t feasible with a traditional server approach.
Networks that offer this kind of functionality typically use a reverse connection model—instead of waiting for inbound connections, your proxy node initiates an outbound connection to a central relay server, maintaining a persistent tunnel. This allows external clients to route traffic through your proxy without requiring open ports on your router.
To implement something similar:
Use reverse proxy tunneling techniques such as reverse SSH tunnels or tunnel services that create outbound connections from your machine and expose them via a public URL or port.
Build or integrate with a custom relay system where each proxy client connects out to a central hub, which then forwards traffic back and forth.
In short, to avoid port forwarding, the key is to reverse the connection direction — have your proxy client connect out, not listen in.
Also, if you're focused on reliability and residential IP quality, looking into the Best Residential Proxies can help improve performance and success rates for your use case.
I have the same requirement. Were you able to find a working solution? Specifically, I'm looking to enforce a Conditional Access (CA) policy only for SharePoint and OneDrive without impacting other services like Teams.
Let me know if you were able to achieve this successfully.
Reasons:
RegEx Blacklisted phrase (1): Were you able to find a
#!/bin/bash
set -euo pipefail
usage() {
echo "Usage: $0 -u <clone_url> -d <target_dir> -b <branch>"
exit 1
}
while getopts "su:d:b:" opt; do
case $opt in
u) CLONE_URL=$OPTARG ;;
d) TARGET_DIR=$OPTARG ;;
b) BRANCH=$OPTARG ;;
*) usage ;;
esac
done
if [[ -z "${CLONE_URL-}" || -z "${TARGET_DIR-}" || -z "${BRANCH-}" ]]; then
usage
fi
git clone --filter=blob:none --no-checkout "$CLONE_URL" "$TARGET_DIR"
cd "$TARGET_DIR"
git config core.sparseCheckout true
{
echo "/*"
echo "!/*/"
} > .git/info/sparse-checkout
git checkout "$BRANCH"
git ls-tree -r -d --name-only HEAD | xargs -I{} mkdir -p "{}"
exit 0
explanation
This script performs a sparse checkout with the following behavior:
1. Clone the repository without downloading file contents (--filter=blob:none) and without checking out files initially (--no-checkout).
2. Enable sparse checkout mode in the cloned repository.
3. Set sparse checkout rules to:
Include all files at the root level (/*).
Exclude all directories at the root level (!/*/).
4. Checkout the specified branch, applying the sparse checkout rules. Only root-level files appear in the working directory.
5. Create empty directories locally by git ls-tree -r -d --name-only HEAD listing all directories in the repo and making those folders. This recreates the directory structure without file contents because Git does not track empty directories.
To fully reset Choices.js when reopening a Bootstrap modal, you should destroy and reinitialize the Choices instance each time the modal is shown. This ensures no cached state or UI artifacts persist:
javascript
$('#customTourModal').on('show.bs.modal', function () {
const selectors = [
'select[name="GSMCountryCode"]',
'select[name="NumberOfAdult"]',
'select[name="HowDidYouFindUsID"]'
];
selectors.forEach(sel => {
const el = this.querySelector(sel);
if (el) {
if (el._choicesInstance) {
el._choicesInstance.destroy();
}
el._choicesInstance = new Choices(el, {
placeholder: true,
removeItemButton: true,
shouldSort: false
});
}
});
});
This approach ensures the Choices.js UI is reset cleanly every time the modal is reopened.
If you’re also looking for a tool that can convert your images or documents to TIFF format, then you should use the BitRecover TIFF Converter tool. This tool comes with many advanced features, such as bulk mode, which allows you to convert not just a single file but multiple files in bulk at once. There is no data loss during the conversion process. This tool saves both your time and effort, and it makes the entire process much faster.
I have some excellent news for you - my timep bash profiler does exactly what you want - it will give you per-command runtime (both wall-clock time and CPU time / combined user+sys time) and (so long as you pass it the -F flag) will generate a bash native flamegraph for you that shows actual bash commands, code structure, and colors based on runtime.
timep is extremely simple to use - download and source the timep.bash script from the github repo (which loads the `timep function and sets up for using it), and then run
timep -F codeToProfile
And thats it - timep handles everything, no need to change anything in the code you want to profile.
As an example, using timep to profile this test script from the timep repo (by running timep -F timep.tests.bash) gives the following profile:
and generates this flamegraph (that shows both the wall-clock time flamegraph and the CPU-time flamegraph).
Note: stack overflow doesn't support SVG images, so I've converted it to a PNG image below. the SVG image I linked (on github) has tooltips and will zoom in on clicking a box and search and things like that. the best way to ensure all the "extras" work is to download the SVG image and then open the local copy.
I simply stopped using psycopg2 and switched to psycopg (aka psycopg3) and everything worked perfectly. I spent a whole day trying to understand why it kept giving this error, and I came to no conclusion. I tried thousands of things and nothing worked, so I just switched.
PostgreSQL is not designed primarily for heavy linear algebra. Pure PL/pgSQL implementations (like Gauss-Jordan) would be very slow and inefficient for 1000x1000. Extensions are the way to go, but availability and performance vary.
PgEigen is a PostgreSQL extension providing bindings to Eigen C++ linear algebra library.It supports matrix inversion and other matrix operations efficiently. Pros: Fast, tested on large matrices, uses compiled C++ code. Cons: Needs installation of C++ dependencies and admin rights.
OneSparse is specialised for sparse matrices and might not be ideal for dense 1000x1000.
How to attribute backup costs to specific Cloud SQL instances?
When you're tracking costs in Google Cloud, SKU Cloud SQL: Backups in [region] are billed based on usage, but the lack of a resource.id in the billing export makes it tough to tie these costs directly to specific Cloud SQL instances. However, work around would be by Instance naming convention and Using billing API filters.
Instance Naming Convention: While this doesn't appear in the billing export, you can match your billing entries with the Cloud SQL instance names manually. For example, if you have instances like prod-db, dev-db, etc., it can help you identify the backups by relating them to specific environments.
Use Billing API and create custom Filters: Even though resource.id isn’t available, you might be able to filter by SKU (e.g., "Cloud SQL: Backups"), regions, and time ranges to make educated guesses. This might still not give you the exact resource ID, but limiting by the filters can help you break down the cost.
Is there a way to correlate billing lines with instance names or labels?
Unfortunately, the billing export you have doesn’t contain labels or instance IDs, which would normally help tie the cost to specific instances. However, there’s a workaround:
Enable Label-based Billing on Cloud SQL: You can add labels to your Cloud SQL instances. Labels are key-value pairs that allow you to tag resources. Once you add labels (like instance-name or environment: production), you can filter the billing export by those labels and identify which instance is generating the backup costs.
Resource IDs for backups: While resource.id might not appear in your current export, you can try to enable more granular billing tracking for backups by using Cloud Monitoring (formerly Stackdriver) and creating custom reports based on your labels or instance names. This way, you could match metrics to billing costs.
How can I identify if a particular backup is unnecessary or consuming excessive storage?
To track excessive storage or unnecessary backups, it’s all about monitoring and data management.
Cloud SQL Monitoring Metrics: Check the backup_storage_used metric (you mentioned you've already checked it). This can help you identify the trends in storage usage and determine if a particular instance is using significantly more storage than expected.
Here’s what you need to look for and compare the expected size of your backups (based on the size of your databases) with the storage usage reported in the metric. If it’s unusually high, it might indicate that backups are growing unexpectedly due to things like; Unnecessary large data retention, Backup frequency, and Non-incremental backups.
Any tips, tools, or workflows to bridge the gap between backup costs and specific Cloud SQL instances?
Google Cloud Billing Reports: You can explore Google Cloud Cost Management tools, such as Cost Explorer or Reports, to break down costs based on project or label. Though not as granular as having a direct resource ID in the billing export, Cost Explorer helps you track costs over time.
Cloud Monitoring: This tool could set up usage-based alerts for your Cloud SQL instance's backup storage. By correlating Cloud SQL storage metrics (like backup_storage_used and total_backup_storage) with backup events, you can monitor abnormal growth or unnecessary backups.
BigQuery Billing Export: Set up BigQuery exports for your billing data. With BigQuery, you can analyze the billing data more flexibly. You could potentially join billing data with other instance-level data (like Cloud SQL instance IDs or tags) to get a clearer picture of which instance is incurring the backup costs.
Here are some helpful links that may help resolve your issue:
This works because a missing some.value will lead to an empty string which is "falsy" and then the elvis operator goes with the fallback. In the SpEL expression - so in the #{...} scope - null means null and not the string "null".
As compared to @EpicPandaForce's answer:
This does not require a custom PropertySourcesPlaceholderConfigurer
but requires a SpEL expression
and is not global - so each property like that will need the adjustment
Blank values are treated as "missing" too - which may be a pro or a con
Thanks to @woxxom I've been able to get the embedded iframes to load by removing initiatorDomains:[runtimeId] and instead using tabIds:[tabId] and updating session rules instead of dynamic rules:
On a sidenote, I found an unrelated error for my use case that says:
Uncaught SecurityError: Failed to read a named property 'document' from 'Window': Blocked a frame with origin "https://1xbet.whoscored.com" from accessing a cross m-origin frame.
This is the src of the parent iframe embedded in the extension page. I'm not sure if this is something I should worry about.
For me, product reviews were disabled while I was creating the majority of my listings and for some reason the review section didn't appear for these products when I enabled it globally later, even though it did work for products that I've created after enabling it.
Enabling reviews on the edit product page of these products didn't work either, BUT it seemed to work for me when I clicked the "Quick Edit" in my Products page for my product and enabled it there.
My question got downvoted but no reason why, such a shame.
I have solved how to do this. I was going to publish a comprehensive answer but I'm getting more disappointed with downvotes and admins posting unnecessary comments so here's a short answer.
Developing a marketplace web app for the Chrome Web Store is a smart way to reach users directly through their browsers, especially if your platform offers digital tools, services, or extensions. However, to stand out in a competitive environment, your app must be well-designed, secure, and performance-optimized.
At its core, Chrome Web Apps are built using standard web technologies like HTML, CSS, and JavaScript. But when you're building a marketplace, you need to factor in multi-vendor capabilities, user accounts, secure payments, and real-time functionality—all of which require strategic marketplace app development.
Key considerations include integrating Chrome APIs properly, ensuring seamless user authentication (such as OAuth), secure backend connections, and maintaining data privacy. Additionally, your app must comply with Chrome Web Store policies, including HTTPS hosting and content restrictions.
If you're serious about launching a scalable and feature-rich marketplace through the Chrome Web Store, it's essential to work with a team experienced in marketplace app development. They can guide you through best practices, build a strong technical foundation, and ensure your app is optimized for user engagement and future growth.
In short, success in the Chrome ecosystem starts with smart planning and expert development tailored to marketplace dynamics.
I know that the question is related to PHP, but if anyone have any problems with escaping $ you can alternatively wrap it in character list like this: [$]\w+[$] .
Based on the official Python documentation and common implementation details, the expression L[a:b] = L[c:d] does indeed create a new, temporary list for the right-hand side L[c:d] before the assignment to the left-hand side L[a:b].
I get the same problem and I am not using Msys2 but Ada under a version of GNAT Studio and Gtkada from 2021. Searching on internet show as a probable cause the fact that I am using now an high definition screen. Programs that I had compiled earlier also had the problem just by running there exe file.The new one also shows this problem while the compilation crates a .exe file.
If I try the command gap -c Print("Hello"); on my Ubuntu, I get an error. In my case the right syntax seems to be :
gap -c "your_code"
or :
gap -c 'your_code'
Work for me : gap -c 'Print("Hello\n");' (I use ' instead of " for a correct parsing... and the \n matter !).
You can try a more simple thing : gap -c "a:=1;" and check in the console that a is indeed bound and equal to 1. However the field in GAPInfo.CommandLineOptions; for the option -c still be empty, I don't think this is where the data is stored. You can recover your input by calling GAPInfo.InitFiles;.
To sum up there is a screen of running the following command :
gap -c 'a:=1; Print("ThisOneIsDisplayed\n"); Print("ThisOneNot");'
